Taming the Beast: EMR Personalization for Internal Medicine Residents

Poster Number: P01
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Teaching Innovation, Curriculum Development, Workflow, Documentation Burden, Usability, Human-computer Interaction, Education and Training, Educational Collaboration
Primary Track: Applications
Programmatic Theme: Academic Informatics / LIEAF

While clinicians who personalize their EMR workflows have improved efficiency and satisfaction, personalization techniques are underutilized. A one-hour EMR personalization workshop was delivered to internal medicine residents. Signal, an EMR audit-log database, tracked personalization metrics of 26 workshop-participants and 49 non-participants from July through November 2023. Participants had a significant increase in Proficiency Score of 1.28 (p<.01, 95% CI [0.69, 1.88]) and non-participants did not. EMR personalization workshops can augment residency training curricula.

Speaker(s):
Jared Silberlust, MD MPH
NewYork-Presbyterian-Weill Cornell Medical Center

Author(s):
John Travis Gossey, MD - Weill Cornell Medical College;

FHIR®-ing up Clinical Research: Outcomes from an Innovative NIH Training Initiative

Poster Number: P02
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Education and Training, Curriculum Development, Data Standards
Primary Track: Foundations
Programmatic Theme: Academic Informatics / LIEAF

Inconsistent integration of HL7® FHIR® into clinical research is partly due to challenges like limited tailored trainings for this community. NIH developed and delivered customized FHIR trainings for researchers within IDeA-CTR institutions. A capability assessment revealed varying levels of FHIR readiness. Hands-on workshops, covering data extraction and analysis with FHIR Bulk Data, SMART on FHIR, enhanced participants’ understanding but underscored additional training needs. To broaden training reach, a train-the-trainer module was incorporated.

Speaker(s):
Snipta Mallick, BS in Computer Science, BS in Cognitive Science
The National Institutes of Health

Author(s):
Steve Tsang; Belinda Seto, PhD - National Institutes of Health; Max Masnick, PhD; Nichole Persing;

Growth of AMIA First Look Program: Lessons Learned and Sustainability

Poster Number: P03
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Diversity, Equity, Inclusion, Accessibility, and Health Equity, Educational Collaboration, Education and Training
Primary Track: Applications
Programmatic Theme: Academic Informatics / LIEAF

The AMIA First Look formed in 2017 is a pathways program for introducing undergraduate women in STEM to informatics and prioritizes underrepresented students. For 2022 and 2023, the program was expanded using multi-pronged approaches with additional funding, inclusive language and collaborations. The 2022 program had 41 students who attended, the highest in 7 years. Feedback from First Look participants has been very positive. Discussions with First Look committee summarized the lessons learned and sustainability strategies.

Speaker(s):
Sripriya Rajamani, MBBS, MPH, PhD, FAMIA
None

Author(s):
Karmen Williams, DrPH, MBA, MSPH, MA, CPH - City University of New York; Young Lee, PhD - University of Pittsburgh; Victoria Tiase, PhD, RN, NI-BC, FAMIA, FAAN, FNAP - University of Utah; Suzanne Boren, PhD, MHA, FACMI, FAMIA - University of Missouri; Tiffani Bright, PhD - Cedars-Sinai Medical Center;

A minimum metadata approach to sharing computable phenotypes across networks and communities

Poster Number: P04
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Standards, Reproducibility, Data Sharing
Primary Track: Applications

Metadata is essential to making computable phenotypes shareable, and Findable, Accessible, Interoperable and Reusable (FAIR). When computable phenotypes are more shareable, they are more easily reused, with implications for research, as well as clinical and public health applications. Given the diverse needs of different research communities and networks, there will likely never be consensus on a single metadata schema for computable phenotypes. However, a common minimum set of metadata could help to make computable phenotypes shareable across research groups and networks, while recognizing the unique needs of different use cases and communities of practice. As a first step towards identifying a set of minimum metadata for computable phenotypes, we extracted and combined the metadata elements described in three recent papers and analyzed the diverse approaches. This poster will describe the similarities and differences between the three approaches, and propose a minimum set of metadata. This is pragmatic first attempt to discover what metadata are most valuable to developers, users, and stewards of computable phenotypes.

Speaker(s):
Marisa Conte, MLIS
University of Michigan

Author(s):
Marisa Conte, MLIS - University of Michigan; Rachel Richesson, PhD, MPH, FACMI - University of Michigan Medical School;

Enhancing Graduate Student Learning Experience on Standards: Collaborative Learning and Mapping to the OMOP Common Data Model

Poster Number: P05
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Education and Training, Educational Collaboration, Data Standards, Knowledge Representation and Information Modeling
Primary Track: Applications
Programmatic Theme: Academic Informatics / LIEAF

Multiple national initiatives have harmonized electronic health record (EHR) clinical data in the form of “flowsheets”. Learning to map flowsheet data to standardized terminology like the Observational Medical Outcomes Partnership (OMOP) offers an opportunity to develop health informatics skills. This activity introduced Doctor of Nursing Practice (DNP) Nursing Informatics (NI) students to mapping methodology, evaluate the standardized terminology for representativeness of nursing flowsheet data, and develop training tools for future learners.

Speaker(s):
Robin Austin
University of Minnesota, School of Nursing

Author(s):
Elizabeth Umberfield, PhD, RN - Mayo Clinic; Sripriya Rajamani, MBBS, MPH, PhD, FAMIA - None;

Generating Multiwords from Verb Complementation Types in the Lexicon

Poster Number: P06
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Natural Language Processing, Information Retrieval, Data Mining
Primary Track: Applications
Programmatic Theme: Academic Informatics / LIEAF

Multiwords are words that contain space(s), such as “tear break up time”. They are recorded as base forms in the SPECIALIST Lexicon. Multiwords from light verb constructions (e.g. “give birth”) and verb-particle constructions (e.g. “tear down”) are encoded in verb complementation types in Lexical records of “give” and “tear”, respectively. Thus, they are not characterized as multiwords in the Lexicon. This work amends this gap. The derived multiwords are scheduled to be released in Lexicon.2025.

Speaker(s):
Chris Lu
NIH/NLM/MSC

Author(s):
Amanda Payne, PhD - NIH/NLM/ACIB; Anna Ripple - National Library of Medicine; James Mork, Master of Science - NIH/NLM/LHNCBC;

Strategies to Introduce Future Pharmacists to Health Information Technology

Poster Number: P08
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Patient Safety, Education and Training, Educational Collaboration, Teaching Innovation, Curriculum Development
Primary Track: Foundations
Programmatic Theme: Academic Informatics / LIEAF

Our objective is to describe teaching strategies used to integrate health IT training into a patient safety course for pharmacy students, and illustrate how health IT training could be incorporated into educational curricula for other HCPs. This presentation will outline at least four distinct teaching strategies for health IT along with at least three lessons learned. This poster presentation can inform educational approaches for equipping a variety of HCPs with foundational knowledge of health IT.

Speaker(s):
Alissa Russ-Jara, PhD
Purdue University

Author(s):
Kyle Hultgren, PharmD - Purdue University; Ephrem Abebe, PhD - Purdue University; Daniel Degnan, PharmD, MS, CPPS, FASHP - Purdue University;

Prompt Engineering with GPT3.5 for Enhancing Information Extraction in Medical Texts

Poster Number: P09
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Information Extraction, Natural Language Processing, Large Language Models (LLMs)
Primary Track: Foundations
Programmatic Theme: Academic Informatics / LIEAF

We explore the viability of employing GPT-3.5 for the extraction of biomedical named entities from biomedical texts dataset by incorporating different prompt-based strategies. The results show an improvement on F1-score over five NER datasets after appropriate prompt engineering improvements. Our findings indicate that utilizing LLMs as a joint source of prior knowledge can be a viable approach for improving the state of the art for few-shot learning-based NER in medical text.

Speaker(s):
Yao Ge, Master
Emory University

Author(s):
Sudeshna Das, PhD - Emory University; Abeed Sarker, PhD - Emory University School of Medicine;

Does Autotext Usage Decrease Documentation Time Among Resident Physicians?

Poster Number: P10
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Documentation Burden, Workflow, User-centered Design Methods, Usability, Informatics Implementation, Education and Training
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Little is known about autotext usage among resident physicians or in inpatient settings. We studied the association between autotext usage and documentation time among resident physicians at a large academic medical center. After controlling for relevant confounders, autotext adoption was associated with decreased documentation time, but among autotext adopters there was no overall association between autotext usage and documentation time. These findings differ significantly from past studies focused on attending physicians in ambulatory settings.

Speaker(s):
Noah Stanco, MD
University at Buffalo Jacobs School of Medicine and Biomedical Sciences

Author(s):
Noah Stanco, MD - University at Buffalo Jacobs School of Medicine and Biomedical Sciences; Samuel Tiosano, MD, MPH - University at Buffalo; Randeep Badwal, MD - University at Buffalo School of Medicine and Biomedical Sciences; William Kelly, MD - University at Buffalo School of Medicine and Biomedical Sciences; Michele Lauria, MD - Kaleida Health;

Comprehensive Self Service Mortality and Readmission Dashboards

Poster Number: P11
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Transitions of Care, Healthcare Quality, Information Visualization, Data Sharing, Data Transformation/ETL
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

Understanding hospital mortality and readmission data is a critical priority for most hospitals, as these outcomes have a considerable impact on reimbursement, CMS star ratings, and brand image. We present Tableau dashboards we have developed using an iterative design process from standardized internal and peer-reviewed benchmark data. These dashboards have enabled understanding of our mortality and readmission data, empowering Geisinger to better prioritize and act on high-value opportunities.

Speaker(s):
Eric Reich, MSHI
Geisinger

Author(s):
Biplab S Bhattacharya, PhD - Geisinger Health System; David Vawdrey, PhD - Geisinger; Hao Liu, MA - Geisinger; Amy Minnich, RN, MHSA - Geisinger; Casey Cauthorn, MIE - Geisinger; Michael Ellison, MS - Geisinger;

Impact of Technostress and Nursing Informatics Competence on Nursing Work Performance in South Korea

Poster Number: P12
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Nursing Informatics, Clinical Decision Support, Healthcare Quality, Patient Safety
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

Background: As the rapid advancement of information technology in clinical settings continues, the significance of nurses' informatics competency in enhancing nursing work performance is increasingly emerging1. This competency allows nurses to make informed decisions confidently, directly improving patient care quality2. Conversely, insufficient informatics skills can induce technostress (e.g., A negative psychological state that occurs when technology users are unable to keep up with new technologies due to the development of information technology)3. With the fast-paced changes in technology, nurses may encounter challenges in acquiring information and experience technostress while developing work-related skills. Despite the clear link between informatics competency, technostress, and nursing performance, there is a gap in research exploring these relationships.

Speaker(s):
Gyuli Baek, Nursing/Doctor
University of pittsburgh

Author(s):
Young Lee, PhD - University of Pittsburgh; Eunju Lee, Dr - Keimyung University College of Nursing;

Health Related Quality of Life of All of Us Young Adults with Chronic Illness

Poster Number: P13
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Patient / Person Generated Health Data (Patient Reported Outcomes), Personal Health Informatics, Transitions of Care, Healthcare Quality, Disability, Accessibility, and Human Function, Nursing Informatics
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Young adulthood is a crucial developmental stage potentially complicated by severe chronic illness. Data from the All of Us Research Program were used to analyze physical and mental health related quality of life in this population without further burdening an already taxed population. Higher rates of fair/poor physical health were reported than fair/poor mental health. This work can inform future studies aimed at improving care for young adults living with severe chronic illness.

Speaker(s):
Carolina Gustafson, PhD
University of Pittsburgh

Author(s):
Theresa Koleck, PhD, BSN, RN - University of Pittsburgh;

Assessing the need for a clinical decision support tool to improve random urine drug screening for patients on chronic opioid therapy

Poster Number: P14
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Clinical Guidelines, Patient Safety
Primary Track: Applications
Programmatic Theme: Clinical Informatics

We assessed the necessity of a clinical decision support (CDS) tool to aid providers in scheduling random urine drug screens (UDS) for patients on chronic opioid therapy (COT) to mitigate opioid overdose risk. Time intervals between screenings often exceeded recommended guidelines and increased over time. Less than one-third of COT patients received UDS as per guidelines, indicating a need for a CDS tool to facilitate proper ordering of random UDS.

Speaker(s):
Hyunjoon Lee, MS
Vanderbilt University Department of Biomedical Informatics

Author(s):
Hyunjoon Lee, MS - Vanderbilt University Department of Biomedical Informatics; Julie Kim, PharmD - Department of Veterans Affairs; Taylor W. Butler, PharmD - Vanderbilt University Medical Center; Lei Wang, MPhil - Vanderbilt University Medical Center; Allison McCoy, PhD, ACHIP, FACMI, FAMIA - Vanderbilt University Medical Center; Heather Jackson, PhD - Vanderbilt University Medical Center; Tyler W. Barrett, MD - Vanderbilt University Medical Center; Daniel B. Larach, MD - Vanderbilt University Medical Center; Hayley H. Rector, PharmD - Vanderbilt University Medical Center; Scott Nelson, PharmD, MS, FAMIA, ACHIP - Vanderbilt University Medical Center;

Preparing for Generative AI: Identifying Common EHR Challenges Contributing to Documentation Burden for Nurses

Poster Number: P15
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Documentation Burden, Nursing Informatics, Qualitative Methods
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

This study aimed to identify common EHR challenges that mental health nurses experience including increasing documentation requirements and inefficient EHR processes. This study leverages nursing engagement to understand their experiences and provides further insight about the use of technology to improve the efficiency of EHR systems and documentation. EHR system challenges, improvements, and the potential use of generative AI will be discussed as a way to better support nursing staff.

Speaker(s):
Jessica Kemp, MHI
Centre for Addiction and Mental Health (CAMH)

Author(s):
Charlotte Pape, MHI - CAMH; Danielle Shin, RN MScN - CAMH; Sara Ling, PhD; Gillian Strudwick, RN, PhD - Centre for Addiction and Mental Health;

“Get out of my way ICU screen!” Needs Assessment for Adapting an Established Pneumonia CDSS for an Interoperable Platform

Poster Number: P16
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Human-computer Interaction, User-centered Design Methods
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Clinical decision support systems (CDSS) improve outcomes for patients with pneumonia but are not widely used due to sub-optimal usability and siloing within EHR platforms. To adapt an existing “ePneumonia” CDSS to a SMART-on-FHIR interface, we interviewed 5 ED clinicians to identify user needs for redesign. We identified design features that conflicted with a discontinuous and fragmented workflow. CDSS for pneumonia in the ED needs to accommodate non-linear navigation while alerting deviation from guideline-recommended care.

Speaker(s):
Matthew Christensen, MD
Vanderbilt University Medical Center

Author(s):
Matthew Christensen, MD - Vanderbilt University Medical Center; Russ Beebe, BA - Vanderbilt University Medical Center; Kathryn Kuttler, PhD - Intermountain Health; Jason Carr, MD - Intermountain Medical Center; Carrie Reale, MSN, RN-BC - Vanderbilt University Medical Center; Michael Ward, MD, PhD, MBA - Vanderbilt University Medical Center; Shilo Anders, PhD - Vanderbilt University Medical Center;

Enhancing Patient-Centered Diabetes Care: Simulated Assessment of SEE-Diabetes, an Educational Decision Support System

Poster Number: P17
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Evaluation, Patient Engagement and Preferences, User-centered Design Methods
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Diabetes affects a substantial portion of the elderly population in the United States, with a significant gap observed in accessing diabetes self-management education and support (DSMES) within the first year of diagnosis. Previous studies have highlighted the lack of patient-centered DSMES during clinic visits, prompting the development of SEE-Diabetes (Support-Engage-Empower-Diabetes), a patient-centered educational tool tailored for older adults with diabetes. SEE-Diabetes aims to aid healthcare providers in delivering DSMES effectively, promoting patient involvement, and encouraging shared decision-making. Implemented in simulated clinical encounters at the University of Missouri Health Care, this feasibility study assessed SEE-Diabetes's acceptance, usability and preliminary efficacy in fostering patient engagement and collaborative goal setting.

The study involved 12 clinical encounters captured on video, including three simulated patients and four providers, analyzed using a mixed-methods approach. Results showed that active engagement with SEE-Diabetes by healthcare providers significantly improved shared decision-making and facilitated patient-centered goal setting. Providers who actively utilized SEE-Diabetes demonstrated higher scores on validated measures of patient involvement and shared decision-making compared to those who used it less frequently or solely for documentation.

Feedback from providers indicated positive perceptions of SEE-Diabetes, highlighting its concise, patient-centered approach. The study underscores the potential of SEE-Diabetes in enhancing patient-provider interactions and empowering patients to take a more active role in their healthcare decisions. Future plans involve further refinement of SEE-Diabetes and additional studies to evaluate its usability, feasibility, and clinical efficacy in real-world clinical settings across primary care and specialty care environments.

Speaker(s):
Min Soon Kim, PhD
University of Missouri

Author(s):
Min Soon Kim, PhD - University of Missouri; Uzma Khan, MD - University of Missouri-Columbia; Margaret Day, MD - University of Missouri - Columbia; Ploypun Narindrarangkura, MD, MS, PhD - Phramongkutklao College of Medicine; Siroj Dejhansathit, MD - University of Missouri; Eduardo Simoes - University of Missouri; Suzanne Boren, PhD, MHA, FACMI, FAMIA - University of Missouri;

Comparison and Cost Strategies Using AI Chat vs API to Map CDEs Across Studies of Chronic Diseases and Disparities in the Health Equity Action Network (HEAN) Research Consortium

Poster Number: P18
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Data Sharing, Knowledge Representation and Information Modeling, Health Equity, Natural Language Processing, Population Health, Informatics Implementation, Racial Disparities
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

HEAN is a national consortium for the prevention and management of multiple chronic conditions (46 R01-level studies). The Research Coordination Center's role is to harmonize common data elements (CDEs) (>8K question items) across studies. We piloted Versa (chat and API), UCSF’s secure generative AI platform as a mapping tool, on a subset of 5 studies (1,304 CDE items) and assessed mapping accuracy, time, and cost. We will discuss results, strategies, and recommendations.

Speaker(s):
Hyelee Kim, MD, MAS, MS
UCSF

Author(s):
Shuang Liang, MS - University of California San Francisco; Kathy Lanier, MPH - University of California San Francisco; Sarit Helman, MPH - University of California San Francisco; Stuart Gansky, DrPH - UCSF; William Brown III, PhD, DrPH, MA - Department of Medicine and Epidemiology & Biostatistics, University of California San Francisco;

Identification of Fall Risks in Home for Older Adults: Image-to-Text Analysis

Poster Number: P19
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Nursing Informatics, Data Mining, Patient Safety
Working Group: Nursing Informatics Working Group
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This study aims to fill the gap in fall risk assessments among the elderly by leveraging artificial intelligence technologies to conduct a comprehensive home environment assessment. A methodological pipeline is proposed that integrates image-to-text and natural language processing algorithms to identify fall risk factors from medical images and textual descriptions.

Speaker(s):
Jiyoun Song, PhD
University of Pennsylvania School of Nursing

Author(s):
Min Jeoung Kang, PhD - Brigham and Women's Hospital/ Harvard Medical School; Boeun Kim, PhD - Johns Hopkins University School of Nursing; Wonkyung Jung, PhD - Johns Hopkins University School of Nursing;

Mapping Stakeholders for Human Trafficking Technology-based Interventions

Poster Number: P20
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Delivering Health Information and Knowledge to the Public, Patient Engagement and Preferences, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Informatics Implementation
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This work aims to identify potential stakeholders to consider when designing anti-trafficking interventions in healthcare contexts, and their potential role as co-creators of future interventions, and participants in studies. We also identify potential barriers and solutions to engagement. We identified 8 stakeholder groups that could inform future interventions. While each group brings meaningful insights, further research is needed on multi-stakeholder engagement strategies to enhance dialog among groups and roles.

Speaker(s):
Michelle Gomez
Vanderbilt University

Author(s):
Michelle Gomez - Vanderbilt University; Kim Unertl, PhD - Vanderbilt University Medical Center;

Implementing a Preoperative Pediatric Anxiety Scale

Poster Number: P21
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Pediatrics, Surgery
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

We aim to increase valid Child Life consults, and consequently decrease anxiolytic use, for anxious preoperative pediatric patients by implementing a validated pediatric anxiety scale into the electronic health record (EHR). This intervention allowed objective identification of anxious patients, associated with increased Child Life consults. No decrease in anxiolytic use emerged, suspected to be failure to address physician habits. This observation highlights and reinforces the importance of considering user practices in successful CDS solutions.

Speaker(s):
Derek Ngai, MD
UT Southwestern

Author(s):

Improving Nutritional Lab Monitoring in Pediatric Patients with Feeding Tubes

Poster Number: P22
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Pediatrics, Patient Safety
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

Our institution identified inconsistent nutrition lab monitoring in feeding tube dependent patients on home enteral nutrition (HEN). We aimed to improve adherence to standardized annual nutrition lab evaluations for HEN patients. Using clinical decision support tools, appropriate nutritional lab ordering rate improved from 2% to 15% for all patients with feeding tubes. Minimally interruptive reminders and standardized lab panels increased nutritional lab ordering rates in HEN patients, which may improve nutritional outcomes.

Speaker(s):
Derek Ngai, MD
UT Southwestern

Author(s):

Understand Variations in Clinical Performance Measures in Anesthesia Providers to Identify Learning Opportunities for Performance Improvements

Poster Number: P23
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Information Visualization, Education and Training
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This work indicates that variations in clinical performance measures may imply learning opportunities for comprehensive performance feedback evaluations and subsequent performance improvements. Visualizations of such variations (e.g., a heatmap) synthesize otherwise noisy information into readable formats. Results from this work may also be leveraged to identify learning opportunities for professional learning and to guide future work in the development of coaching and appreciation algorithms for the prioritization of precision feedback messages.

Speaker(s):
Gan Shi, PhD
University of Michigan, Ann Arbor

Author(s):

Enhancing Precision in Health Risk Assessment: A Machine Learning Framework with Physician-Verified Patient-Generated Health Data

Poster Number: P24
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Patient / Person Generated Health Data (Patient Reported Outcomes), Clinical Decision Support, Precision Medicine, Machine Learning, Informatics Implementation, Surveys and Needs Analysis
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Health risk assessments are pivotal in preventive care and clinical decision-making. However, the reliance on diverse and often conflicting data sources, particularly Patient-Generated Health Data (PGHD), introduces significant challenges. We propose an innovative framework that employs Machine Learning (ML) with physician verification to enhance the reliability and accuracy of data used in health risk assessments, optimizing clinical decision support systems (CDSS). Our framework utilizes a supervised ML model, specifically a Random Forest algorithm, to evaluate patient trust and assimilate expert knowledge in determining source credibility. The model, trained on extensive demographic and historical survey data, assigns probability weights to each data source, reflecting its reliability. A subsequent verification step by a primary care physician adjusts the source trust probability, thereby refining the accuracy of the overall risk assessment. This methodology enables dynamic learning from physician inputs and historical data, improving the framework's predictive capability. Our approach promises greater efficiency in clinical decision-making by improving the quality of patient data utilized in CDSS. Prospective developments include expanding the model to embrace more diverse data types and aligning with Fast Healthcare Interoperability Resources (FHIR) standards to ensure broad CDSS interoperability. Future clinical validation will determine the framework's practical application, aiming to solidify its role in precision medicine initiatives. This abstract details the framework's methodology and anticipated impact on healthcare delivery, underscoring the potential of ML in enhancing CDSS through more reliable PGHD integration.

Speaker(s):
Abdulrahman Alsheikh, MD, MPH, PMP
Johns Hopkins School of Medicine - Division of Health Sciences Informatics

Author(s):
Shatha Alshaibi, B.Sc, MPH - Lean Business Services;

Enhancing diagnostic excellence: an eCQM for inpatient pneumonia care

Poster Number: P25
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Clinical Guidelines, Standards
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Electronic Clinical Quality Measures (eCQMs) support quality improvement measurement and reporting. Prior eCQMs for pneumonia focused on treatment, but may have resulted in overdiagnosis. We combined discharge diagnoses codes with natural language processing of chest imaging reports to develop and evaluate an eCQM for pneumonia diagnosis in hospitalized patients, mapped to standardized terminologies. The automated methods could minimize need for manual chart review.

Speaker(s):
Brittany Kent, MS in Biomedical Informatics
University of Utah

Author(s):
Brittany Kent, MS in Biomedical Informatics - University of Utah; Lindsay Visnovsky, PhD, MS - University of Utah; Alec Chapman, MS - University of Utah; McKenna Nevers, MS - University of Utah; Jian Ying, PhD - University of Utah; Katherine Sward, PhD - University of Utah; Barbara Jones, MD MS - University of Utah / Salt Lake City VA;

Hidden Flaws Behind Expert-Level Accuracy of GPT-4 Vision in Medicine

Poster Number: P26
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Imaging Informatics, Deep Learning
Primary Track: Applications

We conducted a comprehensive evaluation of GPT-4V’s rationales when solving NEJM Image Challenges. We show that GPT-4V achieves promising results compared to expert physicians regarding multi-choice accuracy (88.0% vs. 77.0%). However, we discovered that GPT-4V frequently presents flawed rationales even in cases where it makes the correct final choices (27.3%), mostly in image comprehension. As such, our findings emphasize the necessity for in-depth evaluations before integrating such models into clinical workflows.

Speaker(s):
Qiao Jin, M.D.
National Institutes of Health

Author(s):
Qiao Jin, M.D. - National Institutes of Health; Fangyuan Chen, B.S. - University of Pittsburgh; Yiliang Zhou, Master - Weill Cornell Medicine; Ziyang Xu, MD, PhD - New York University Grossman School of Medicine; Justin Cheung, MD - Harvard Medical School and Massachusetts General Hospital; Robert Chen, MD - Weill Cornell Medicine; Ronald Summers, MD, PhD - National Institutes of Health; Justin Rousseau, MD, MMSc - University of Texas Southwestern Medical Center; Peiyun Ni, MD - Harvard Medical School and Massachusetts General Hospital; Marc Landsman, MD - Case Western Reserve University School of Medicine; Sally Baxter, MD, MSc - University of California - San Diego; Subhi Al'Aref, MD - University of Arkansas for Medical Sciences; Yijia Li, MD - University of Pittsburgh Medical Center; Michael Chiang, MD - National Institutes of Health, National Eye Institute; Yifan Peng, PhD - Weill Cornell Medicine; Dept of Population Health Sciences; Div of Health Informatics; Zhiyong Lu, PhD - National Library of Medicine, NIH;

Physician Electronic Health Record Use Surrounding Paid Time Off

Poster Number: P27
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Internal Medicine or Medical Subspecialty, Human-computer Interaction, Documentation Burden
Primary Track: Foundations

This retrospective observational study of 57 primary care physicians examines electronic health record (EHR) use on the first day before or after periods of time away from work, or “buffer workdays”. Using scheduling data and EHR use metrics, we compare the workload on buffer workdays against standard workdays while adjusting for provider variables. Our results indicate that physicians may experience increased in-basket related tasks during buffer workdays.

Speaker(s):
Corey Obermiller, MAS Applied Statistics
Wake Forest School of Medicine

Author(s):
Corey Obermiller, MAS Applied Statistics - Wake Forest School of Medicine; Richa Bundy, MPH - Wake Forest Baptist Health; Lauren Witek, MStat - Atrium Health Wake Forest Baptist; Adam Moses, MHA - Wake Forest Baptist Medical Center; Brad Rowland, MD - Atrium Health Wake Forest Baptist; Lindsey Carlasare, MBA - American Medical Association; Gary Rosenthal, MD, FACP - Wake Forest School of Medicine; Christine Sinsky, MD - American Medical Association; Ajay Dharod, MD - Wake Forest University School of Medicine;

ProvenCare Initiative: Enhancing Patient Outcomes and Healthcare Efficiency

Poster Number: P29
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Information Visualization, Healthcare Economics/Cost of Care, Healthcare Quality
Primary Track: Applications

Geisinger's ProvenCare initiative employs user-centered design and data-driven strategies to improve quality, finance, and resource utilization outcomes. We outline the program’s objective, methodology, and expected impacts, highlighting its potential to promote sustainable healthcare practices using governed internal and external peer benchmark data. We present the iterative dashboard development process, the incorporation of stakeholder input, and illustrate an example of how key performance indicators identified a cost spike in an anesthesia-related medication, within a specific program.

Speaker(s):
Grant Walter, BS
Geisinger

Michelle Dempsey, Bachelors Degree
Geisinger

Author(s):
Michelle Dempsey, BS CPBI - Geisinger; Rachel Van Loan, BS CLSSMBB - Geisinger; Anthony Petrick, MD - Geisinger; Paul Simonelli, MD - Geisinger; Eric Reich, MSHI - Geisinger; David Vawdrey, PhD - Geisinger; Biplab S Bhattacharya, PhD - Geisinger Health System;

Pre-Trained Large Language Models for Food Concept Extraction from Patient-Generated Meal Narratives

Poster Number: P30
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Patient / Person Generated Health Data (Patient Reported Outcomes), Information Extraction, Large Language Models (LLMs), Precision Medicine, Natural Language Processing
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Patient-generated free-text meal records facilitate more informed precision nutrition interventions. However, user generated dietary information is poorly structured and requires costly manual standardization. Named entity recognition (food-NER) is one way to automate this process. However, current food-NER approaches are mainly rule-based and struggle with user-entry errors, abbreviations, and branded food detection. In this study, we show that pre-trained LLMs could potentially improve food-NER and better handle the above challenges for patient-generated data.

Speaker(s):
Yanwei Li, BS
Columbia University

Author(s):
Yanwei Li, BS - Columbia University; Adit Anand, B.S. - Columbia University; Lena Mamykina, PhD - Columbia University; Chunhua Weng, PhD - Columbia University;

Deterioration Phenotypes Through Clustering Analysis of Provider Actions

Poster Number: P31
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Machine Learning, Evaluation, Critical Care, Usability
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This study examines patient deterioration phenotypes by analyzing healthcare provider actions in relation to high Deterioration Index (DTI) scores and then using martingale residuals to perform hierarchical clustering to identify seven distinct deterioration phenotypes. The results demonstrate the potential for more accurate identification of true deterioration, which suggests that recognizing these phenotypes could improve patient outcomes through tailored interventions.

Speaker(s):
Abhinab Kc, B.A.
University of Minnesota Medical School Twin Cities

Author(s):
Abhinab Kc, B.A. - University of Minnesota Medical School Twin Cities; Benjamin Langworthy, PhD - University of Minnesota School of Public Health; Thomas Byrd, MD, MS - University of Minnesota - MHealth Fairview;

Friedman Risk Score: A Custom Primary Care Score Predicting Consumption

Poster Number: P32
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Population Health, Machine Learning, Healthcare Quality
Primary Track: Applications

We extracted data from >100,000 University of California San Diego Health primary care patients to predict top healthcare utilization with six multiclass prediction algorithms. XGBoost outperformed other models, achieving the highest area under the receiver operating characteristic curve of >0.80 for patients with very high healthcare utilization (top 5%). Our findings demonstrate the potential of advanced risk score models to optimize healthcare resource allocation and improve care efficiency.

Speaker(s):
Yufei Yu, BS
University of California San Diego

Author(s):
Yufei Yu, BS - University of California San Diego; Alson Mo, BA - UC San Diego Health; Tsung-Ting Kuo, PhD - University of California San Diego; Amy Sitapati, MD - UC San Diego;

Enhancing Operative Documentation of Cancer Operations: The Transition to Synoptic Operative Reporting at a Cancer Center

Poster Number: P33
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Surgery, User-centered Design Methods, Real-World Evidence Generation, Documentation Burden
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Synoptic operative reports are becoming widely accepted as they have been shown to improve surgeon adherence to higher standards of cancer care. The adoption of synoptic operative notes will further increase as they will be required for accreditation by the American College of Surgeons (ACS) Commission on Cancer (CoC). We demonstrate the feasibility of a large scale implementation of synoptic operative reports for multiple surgical oncology specialties at a free-standing cancer center.

Speaker(s):
Heather Lyu, MD, MBI
University of Texas, MD Anderson Cancer Center

Author(s):
Jesus Gonzalez, BS - MD Anderson Cancer Center; Shelby Holtsmith, BS - MD Anderson Cancer Center; Jonathan Escalante, BA, MBA - MD Anderson Cancer Center; Angela Chandler, MSN - MD Anderson Cancer Center; Stephen Swisher, MD - MD Anderson Cancer Center; Kelly Hunt, MD - MD Anderson Cancer Center; Matthew Katz, MD - MD Anderson Cancer Center; Jolyn Taylor, MD - MD Anderson Cancer Center; SAHIL KAPUR, MD - UNIVERSITY OF TEXAS MD ANDERSON CANCER CENTER;

Dashboard Development for Tracking Performance of High-Risk Abdominal Aortic Aneurysm Screening

Poster Number: P34
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Information Visualization, Population Health, User-centered Design Methods, Machine Learning, Informatics Implementation, Clinical Decision Support
Primary Track: Applications

Geisinger deployed a machine learning model that identifies high-risk individuals for early Abdominal Aortic Aneurysm screening, who are contacted by a nursing team to recommend screening. Program stakeholders wanted to measure the success and impact of the screening program, including screening rates and results. We developed a dashboard in Tableau for use by our clinical and outreach teams to track the progress of identifying, conducting outreach, and screening high-risk patients.

Speaker(s):
Akiva Blickstein
Geisinger

Author(s):
Biplab S Bhattacharya, PhD - Geisinger Health System; Casey Cauthorn, MS - Geisinger; Evan Ryer, MD - Geisinger; Gregory Salzler, MD - Geisinger; Jim Urick, MS - Geisinger Health System; Rebecca Maff, MS; James Elmore, MD - Geisinger; Elliot Mitchell, PhD - Geisinger;

Emergency Department Admitting Service Triage Using Retrieval-Augmented Language Models

Poster Number: P35
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Information Retrieval, Workflow, Informatics Implementation, Internal Medicine or Medical Subspecialty
Primary Track: Applications
Programmatic Theme: Clinical Informatics

We evaluated the utility of large language models (LLMs) for triaging emergency department admissions an academic medical center. Using retrieval-augmented generation (RAG) with existing policy documents, we tested Anthropic's Claude 2 and OpenAI's GPT-4. Our findings show that GPT-4, combined with a departmental spreadsheet, most accurately determined correct admitting services, suggesting a viable method to reduce clerical burden without additional resource investment.

Speaker(s):
Dong-han Yao, MD
Stanford

Author(s):
Dong-han Yao, MD - Stanford; Debadutta Dash, MD - Stanford; Adi Badhwar, MS - Stanford; Wendy Li, MD - Stanford; Nicholas Hall, MD - Stanford; Yoseph Semma, MD - Stanford;

Detecting Duplicates in Adverse Event Reports for Cannabis sativa and Mitragyna speciosa (Kratom) Products

Poster Number: P36
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Drug Discoveries, Repurposing, and Side-effect, Patient Safety, Real-World Evidence Generation, Evaluation, Data Mining
Primary Track: Applications

Duplication of spontaneous safety reports is a concern for botanical pharmacovigilance with few available solutions. We combined probabilistic matching with literature references and adverse event reports details to deduplicate spontaneous reports for cannabis and kratom. Manual review found that the combined approach nearly doubled the identification of duplicates compared to either method alone, but still missed duplicates. We conclude that more research is needed to address the issue of extreme duplication for botanical safety reports.

Speaker(s):
Sanya Taneja, MS
University of Pittsburgh

Author(s):
Sanya Taneja, MS - University of Pittsburgh; Xiaotong Li, MS - University of Pittsburgh; Maryann Chapin, PharmD - University of Pittsburgh; Sandra Kane-Gill, PharmD - University of Pittsburgh; Richard Boyce, PhD - University of Pittsburgh;

“You Can’t Answer Inbox Messaging When You’ve Got a Scalpel in Your Hand”: Health IT-related Barriers to Clinical Workflow in Breast Cancer

Poster Number: P37
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Workflow, User-centered Design Methods, Qualitative Methods, Clinical Decision Support, Human-computer Interaction
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Breast cancer treatment decision-making is a complex process that involves a number preference-sensitive decisions made across a large, multidisciplinary team. To inform the design of a decision support tool for treatment decision-making, we conducted 22 semi-structured interviews with breast cancer clinicians to understand current workflows. We identified numerous barriers to clinical workflows related to the implementation and use of health information technologies. We propose design recommendations to optimize clinical workflow and the use of health IT.

Speaker(s):
Megan Salwei, PhD
Vanderbilt University Medical Center

Author(s):
Megan Salwei, PhD - Vanderbilt University Medical Center; Janelle Faiman, BA - Vanderbilt University Medical Center; Ingrid Meszoely, MD - Vanderbilt University Medical Center; Ben Park, MD, PhD - Vanderbilt University Medical Center; Matthew Weinger, MD, MS - Vanderbilt University Medical Center;

Developing an LLM-based Chatbot using Retrieval Augmented Generation for Families Affected by Complex Lymphatic Anomalies

Poster Number: P38
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Natural Language Processing, Delivering Health Information and Knowledge to the Public, Pediatrics
Primary Track: Applications
Programmatic Theme: Clinical Informatics

We developed a GPT-based chatbot using retrieval-augmented generation (RAG) for complex lymphatic anomalies (CLAs) families. We evaluated four GPT-based models: ChatGPT, GPT (No RAG), GPT (RAG with 1 document), and GPT (RAG with 10 documents) using manual scoring on accuracy and comprehensiveness by a clinical expert. The preliminary results suggest that the CLA chatbot can generate more accurate and understandable answers when incorporating an LLM with a rare disease knowledge base.

Speaker(s):
Min Zhao
Washington University in St. Louis, School of Medicine

Author(s):
Ethan Hillis; Inez Oh, PhD - Institute for Informatics at Washington University in St. Louis, School of Medicine; Aditi Gupta - Washington University in St. Louis; Albert Lai, PhD, FACMI, FAMIA - Washington University; Bryan A. Sisk, MD, MSCI - Washington University in St Louis School of Medicine;

Protocol for a Prediction Model for Late-term Preeclampsia in Pregnant People Without Early Pregnancy Biomarkers

Poster Number: P39
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Racial Disparities, Pediatrics
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Preeclampsia affects long-term health and disproportionately affects US-born Black and/or African American women compared to white women. Preeclampsia can be diagnosed early, but Black women are less likely to access early prenatal care. Current preeclampsia prediction models presume access to early pregnancy data. This poster describes a study that uses shallow machine learning methods for a predictive model using late pregnancy data to improve diagnosis of preeclampsia in mothers who present late for prenatal care.

Speaker(s):
Jill Inderstrodt, PhD/MPH
Indiana University Fairbanks School of Public Health

Author(s):
Kedir Turi, PhD - Indiana University School of Public Health;

Using Large Language Models to Improve Patient-Provider Communications

Poster Number: P40
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Clinical Decision Support, Documentation Burden
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Patient messages often lack important details, resulting in multiple rounds of messages for clinicians to gather the necessary information. We used a fine-tuned large language model (LLM) and ChatGPT to draft follow-up questions for patients. Five physicians rated them for utility, completeness, conciseness, and clarity. ChatGPT and the fine-tuned LLM could generate more useful and complete questions than those from healthcare providers. Overall, LLM shows great potential for improving patient-provider communication through patient portals.

Speaker(s):
Siru Liu, PhD
Vanderbilt University Medical Center

Author(s):
Siru Liu, PhD - Vanderbilt University Medical Center; Aileen Wright, MD - Vanderbilt; Allison McCoy, PhD, ACHIP, FACMI, FAMIA - Vanderbilt University Medical Center; Sean Huang, MD - Vanderbilt University; Julian Genkins - Stanford Health Care; Josh Peterson, MD, MPH - Vanderbilt University Medical Center; Yaa Kumah-Crystal - Vanderbilt Univeristy Medical Center; William Martinez, MD, MS - Vanderbilt University Medical Center; Babatunde Carew, MD - Vanderbilt University Medical Center; Dara Mize, MD, MS - Vanderbilt University Medical Center; Bryan Steitz, PhD - Vanderbilt University Medical Center; Adam Wright, PhD - Vanderbilt University Medical Center;

Implementing a Clinical Decision Support Tool to Reduce Hospital Length-of-Stay for Pediatric Cancer Patients with Low-Risk Fever and Neutropenia

Poster Number: P41
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Clinical Guidelines, Pediatrics, Informatics Implementation, Patient Safety
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Pediatric cancer patients with fever and neutropenia (F&N) are traditionally hospitalized until neutropenia and fever have resolved, regardless of their individualized risk for sepsis, resulting in excessive lengths of stay (LOS) and intravenous antibiotic exposure for low risk patients. We describe the design, implementation and evaluation of a novel clinical decision support tool within the Electronic Health Record to identify low risk patients eligible for early discharge. Early results suggest feasibility, user acceptability, and safety of this approach, with ongoing evaluation needed to confirm the tools' effectiveness.

Speaker(s):
Claire Stokes, MD, MPH
Children's Healthcare of Atlanta and Emory University

Author(s):
Claire Stokes, MD, MPH - Children's Healthcare of Atlanta and Emory University; Wayne Liang, MD MS FAMIA - Children's Healthcare of Atlanta & Emory University; Edwin Ray, RN - Children's Healthcare of Atlanta; Swaminathan Kandaswamy, PhD - Emory University School of Medicine; Evan Orenstein, MD - Childrenís Healthcare of Atlanta; Julia Yarahuan, MD - Children's Healthcare of Atlanta/Emory University;

Validating a Performance Measurement Framework through Real-World Experience in PC CDS Measurement

Poster Number: P42
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Patient / Person Generated Health Data (Patient Reported Outcomes), Patient Engagement and Preferences, Evaluation, Healthcare Quality, Real-World Evidence Generation
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This poster describes the assessment of 20 real-world patient-centered clinical decision support (PC CDS) projects to validate a performance measurement framework and identify future areas of growth for PC CDS measurement. Findings from the document review and interviews revealed ways to advance the measurement of the safety, timeliness, effectiveness, efficiency, equity, and patient-centeredness of PC CDS technologies. These findings provide important focus areas for future monitoring of PC CDS in the age of artificial intelligence.

Speaker(s):
Prashila Dullabh, MD
NORC at the University of Chicago

Author(s):
Prashila Dullabh, MD - NORC at the University of Chicago; Courtney Zott, MPH; Nicole Gauthreaux, MPH - NORC at the University of Chicago; James Swiger, MBE - AHRQ; Edwin Lomotan, MD - AHRQ; Dean Sittig, PhD - University of Texas Health Science Center at Houston;

Leveraging EHR Data to Improve Care Timeliness for Patients with Cardiovascular Implantable Electronic Devices

Poster Number: P43
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Patient Safety, Informatics Implementation
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Cardiovascular Implantable Electronic Devices (CIEDs) like pacemakers and implantable cardioverter defibrillators (ICDs) need to be checked routinely to make sure they are serving patients properly. Our findings showed that a vast majority of the patients with these devices were overdue for an in-person device check. To correct this issue, we used clinical decision support tools to notify clinicians when patients are overdue for an in-person device check.

Speaker(s):
Donald Sengstack, MS


Author(s):
Donald Sengstack, MS; Aarti Dalal, DO - Vanderbilt University Medical Center; Allison McCoy, PhD, ACHIP, FACMI, FAMIA - Vanderbilt University Medical Center;

An Alternative Metric for EHR Data Completeness

Poster Number: P44
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Standards, Precision Medicine, Bioinformatics
Primary Track: Applications

The utility of electronic health record (EHR) data for secondary research can be limited by data missingness, which can arise from medical care received outside of compatible health data systems. We quantified expected EHR frequency using records from a reference cohort of participants with the most complete EHR data, allowing for an estimation of EHR data completeness for the full cohort.

Speaker(s):
John Giannini, Ph.D.
National Institutes of Health

Author(s):
Yechiam Ostchega; Matthew Spotnitz, MD - National Institutes of Health; Emily Clark, MPH - Leidos; Lakshmi Anandan, MPH - Leidos; Lew Berman, PhD - National Institutes of Health;

Development of a Prototype for Patient Artificial Intelligence Guided E-messages (PAIGE)

Poster Number: P45
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Clinical Decision Support, Documentation Burden
Primary Track: Applications
Programmatic Theme: Clinical Informatics

We conducted a feasibility study of constructing a conversational agent prototype embedded within a patient portal to display follow-up questions, seek patient responses for additional context to improve effective communication between patients and clinicians. Using input from 6 physicians about 7 primary care scenarios, we generated follow-up questions from a local fine-tuned large language model and GPT-4. The prototype of Patient Artificial Intelligence Guided E-messages (PAIGE) will be evaluated to assess patient and provider impact.

Speaker(s):
Siru Liu, PhD
Vanderbilt University Medical Center

Author(s):
Siru Liu, PhD - Vanderbilt University Medical Center; Allison McCoy, PhD, ACHIP, FACMI, FAMIA - Vanderbilt University Medical Center; Laura Zahn, MS; Elise Russo - Vanderbilt University Medical Center; Courtney Zott, MPH; Nicole Gauthreaux, MPH - NORC at the University of Chicago; James Swiger, MBE - AHRQ; Edwin Lomotan, MD - AHRQ; Dean Sittig, PhD - University of Texas Health Science Center at Houston; Prashila Dullabh, MD - NORC at the University of Chicago; Adam Wright, PhD - Vanderbilt University Medical Center;

Expert-Augmented Machine Learning for Predicting Extubation in the Pediatric Intensive Care Unit

Poster Number: P46
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Clinical Decision Support, Critical Care
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

Expert-augmented machine learning (EAML) is a novel method that combines machine learning with human expert knowledge to capitalize on and combine the strengths of each. EAML 1) extracts rules from decision-tree machine learning models and 2) subjects them to human expert review, combining these two to create a final, more robust, model. Here, we apply EAML to the prediction of successful extubation in the PICU.

Speaker(s):
Jean Digitale, MPH, RN
UCSF

Author(s):
Deborah Franzon, MD - UCSF; Jin Ge, MD - UCSF; Charles McCulloch, MD - UCSF; Mark Pletcher, MD MPH - UCSF; Efstathios Gennatas, MBBS, PhD - UCSF;

Using Large Language Models for Identification of Patients with Elevated Risk of Breast Cancer

Poster Number: P47
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Natural Language Processing, Cancer Prevention, Information Extraction
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Identifying patients with a family history of breast cancer is an important step in determining high risk individuals who could benefit from BRCA genetic screening. However, identifying eligible candidates for genetic screening based on unstructured notes in electronic health records is labor intensive. Through the use of a large language model, we are able to quickly and accurately identify patients, allowing for the automation of BRCA genetic screening recommendations.

Speaker(s):
Esha Datta, Ph.D
One Medical

Author(s):
Emily Kasa, BS - One Medical; Sergio Martinez-Ortuno, MS - One Medical; Lenny Lesser, MD - One Medical;

Fair Logistic Regression for Intracranial Pressure Monitoring Assignment

Poster Number: P48
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Fairness and Elimination of Bias, Machine Learning, Diversity, Equity, Inclusion, Accessibility, and Health Equity
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This study applies a fairness-constrained formulation and optimization algorithm to construct a fair logistic model for intracranial pressure monitoring. Results show that the fair model improves fairness without sacrificing much of the model’s predictive power as measured by the AUC. Application of this methodology is important for the assessment and mitigation of potential biases in medical care. Future study should further validate and explain the identified bias in gender.

Speaker(s):
Xin Chen, Bachelor of Science
University of California, Berkeley

Author(s):
Malini Mahendra, MD; Anil Aswani, PhD - University of California, Berkeley;

Novel Machine Learning on Asynchronous Clinical Pages to Predict Clinical Deterioration

Poster Number: P49
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Workflow, Clinical Decision Support, Natural Language Processing, Critical Care, Diagnostic Systems, Nursing Informatics
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Current early warning scores for predicting patient deterioration rely on manual EHR data entry around a few limited parameters. We propose a novel machine learning approach using clinical pages, which provide insight into clinicians’ intuition and already part of clinical workflow, to predict deterioration events. Our model detected 53% of events within 24 hours at a specificity of 0.80, outperforming existing methods. This approach can improve patient care, reduce nursing workload, and enhance clinical decision-making.

Speaker(s):
Isabel Arvelo, BA
Vanderbilt University

Author(s):
Bryan Steitz, PhD - Vanderbilt University Medical Center; Adam Wright, PhD - Vanderbilt University Medical Center; Kipp Shipley, DNP - Vanderbilt University Medical Center;

Self-Supervised Learning Methods for Thyroid Nodule Classification

Poster Number: P50
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Deep Learning, Cancer Prevention
Primary Track: Applications

This study explores the utilization of self-supervised learning (SSL) models for the automated classification of thyroid nodules from ultrasound images, given the challenges of accurately interpreting such images and the limited availability of accurately labeled data. Employing self-supervised learning, the research begins with an unlabeled dataset for initial training, followed by fine-tuning with a labeled dataset from Banner-University Medical Center, Phoenix, which includes images from 169 patients. The evaluation leverages architectures like Resnet16, InceptionV3, and DenseNet121, using RotNet to train Convolutional Neural Networks (CNNs) on recognizing image rotations to learn crucial features from unlabeled images, and then finetuning these CNNs on labeled data to accurately assess the necessity of surgery for thyroid conditions. The results demonstrate that SSL, especially when compared with transfer learning (TL) and supervised learning (SL), shows promising performance in the classification of thyroid nodules, indicating that SSL could be a valuable method for handling small-sized datasets in the medical imaging domain.

Speaker(s):
Debottama Das, PhD
University of Arizona

Author(s):

Leveraging Large Language Model to Improve Outpatient Diabetes Management

Poster Number: P51
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Clinical Decision Support, Chronic Care Management
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This study explores the application of the fine-tuned the large language model (ChatGLM) model in outpatient diabetes management and evaluates its effectiveness. Integrated into the hospital information system, it generates personalized treatment recommendations, laboratory tests and medication reminders from patient data. While effective, there are limitations in data entry and complex case handling, indicating a need for further optimization.

Speaker(s):
Jialin Liu, MD
West China Hospital Sichuan University

Author(s):

Quantifying Immediately Released Test Results that Contribute to the Highest Volume of Patient Messages

Poster Number: P52
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Patient Engagement and Preferences, Patient / Person Generated Health Data (Patient Reported Outcomes), Telemedicine, Mobile Health
Primary Track: Policy
Programmatic Theme: Clinical Informatics

This study evaluates the impact of the 21st Century Cures Act on patient messaging volume at Vanderbilt University Medical Center, analyzing 4.8 million test results and 450,679 messages pre- and post-policy implementation. Findings reveal significant increases in messages, especially from sensitive tests previously delayed. Future work will explore message content and test result abnormality, aiming to reduce clinician workload through informed patient communication strategies.

Speaker(s):
Robert Becker, MS
Department of Biomedical Informatics, Vanderbilt University

Author(s):
Robert Becker, MS - Department of Biomedical Informatics, Vanderbilt University; Bryan Steitz, PhD - Vanderbilt University Medical Center; S. Trent Rosenbloom, MD, MPH, FACMI, FAMIA - Vanderbilt University Medical Center Dept of Biomedical Informatics;

MedTitrator: An Open-Source, Generalized Clinical Decision Support Tool for Medication Titration

Poster Number: P53
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Clinical Guidelines, Chronic Care Management
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

A reliable generic framework for medication titration can facilitate implementation of many clinical guidelines into clinical decision support (CDS) tools. We developed an open-source generic framework to quickly design and implement medication titration processes, and used it to build a CDS titration tool for heart failure with reduced ejection fraction. The framework expressed all the required logic reliably and made correct recommendations. This is promising for other use cases but further testing is needed.

Speaker(s):
Saeed Arabi, Resident
University of Utah Health

Author(s):
Saeed Arabi, Resident - University of Utah Health; Kensaku Kawamoto, MD, PhD, MHS - University of Utah;

Real-Time Patient State-Based Optimization Strategies for High-Cost Drug Distribution

Poster Number: P54
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Data Mining, Infectious Diseases and Epidemiology, Healthcare Economics/Cost of Care
Primary Track: Applications

This study develops a demand forecasting model integrating real-time patient status data and machine learning techniques to optimize the distribution of high-cost drugs. We adopted palivizumab, a monoclonal antibody for preventing respiratory syncytial virus (RSV) infections, as a use case. The model demonstrated superior performance compared to traditional forecasting methods, highlighting its potential to revolutionize supply chain management and reduce overall pharmaceutical expenses.

Speaker(s):
Shota Kawamoto


Author(s):
Shota Kawamoto; Yoshihiko Morikawa, MD - Tokyo Metropolitan Children's Medical Center; Yoshihiko Morikawa, MD - Tokyo Metropolitan Children's Medical Center; Hirokazu Sugiyama, PhD - The University of Tokyo; Naohisa Yahagi, MD, PhD - Keio University;

Enhancing Empiric Antibiotic Selection through Personalized Culture Display

Poster Number: P55
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Information Visualization, Patient Safety
Primary Track: Applications
Programmatic Theme: Clinical Informatics

The challenge of selecting appropriate empiric antibiotics for patients with histories of drug-resistant organism (DRO) infections is significant, posing safety risks due to the potential for increased morbidity and mortality from incorrect choices. This study introduces a novel solution designed to enhance clinical decision-making by integrating historical culture data into clinical workflows at the point of culture ordering. By facilitating efficient review and synthesis of past antimicrobial resistance results, the tool aligns with SAFER guidelines to improve initial antibiotic regimen selection. Our approach involved a multidisciplinary team within a quaternary care system, who redesigned workflow elements to streamline culture and medication ordering processes. This included the implementation of an antimicrobial dashboard to facilitate the simultaneous review of multiple results, thereby improving the perception of resistance patterns and efficiency of the ordering process. The tool's efficacy is being assessed by comparing pre- and post-implementation proportions of patient encounters involving inappropriate antibiotic regimens for individuals with DRO histories. Preliminary results indicate a potential reduction in inappropriate empiric antibiotic use among patients with DRO infections, as evidenced by a 2% error rate in antibiotic selection among preliminary patient encounters. The anticipated outcomes include improved antibiotic selection accuracy and enhanced provider efficiency in reviewing and incorporating past culture data. This innovative solution offers a promising approach to preventing inappropriate empiric antibiotic use in patients with DRO histories, potentially averting dangerous medical errors while enhancing clinician efficiency.

Speaker(s):
Aidan Petrovich, MD
Northside Hospital

Aidan Petrovich, MD
Northside Hospital

Author(s):
Aidan Petrovich, MD; Sarah Thompson, MSHIMI, BSN, RN - Children's Healthcare of Atlanta; Swaminathan Kandaswamy, PhD - Emory University School of Medicine; Mark Gonazelz, PhD - Children's Healthcare of Atlanta; Alexis Carter, MD - Children's Healthcare of Atlanta; Preeti Jaggi, MD - Children's Healthcare of Atlanta; Julia Yarahuan, MD - Children's Healthcare of Atlanta/Emory University;

Barriers and Facilitators to Implementing an Interoperable Clinical Decision Support System

Poster Number: P56
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Interoperability and Health Information Exchange, Chronic Care Management, Patient Engagement and Preferences
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This study describes barriers and facilitators to implementing Fast Healthcare Interoperability Resources (FHIR)-based clinical decision support. We analyzed the implementation of two applications in an electronic health record to support chronic pain care. We used data from a multidisciplinary design workshop, interviews with primary care providers and staff, clinic-level workflow analysis, and implementation meeting notes. We identified several barriers that need to be addressed by technology and healthcare organizations interested in implementing interoperable decision support.

Speaker(s):
Jyotsna Gutta, MPH
Center for Health Policy

Author(s):
Jyotsna Gutta, MPH - Center for Health Policy; Magda Francois, MA - University of Florida; julie diiulio, MS; Miranda Reid, MPH - University of Florida; Khoa Nguen, Pharm D - University of Florida; Francisco Martinez-Wittinghan, MD, PhD - UF Health; Lori Bilello; Mario El Hayek, MD; Laura Marcial, PhD - RTI International; Ramzi Salloum, PhD; Christopher Harle, PhD - Indiana University;

Cloud-Based Registry: Advancing CDH Research & Collaboration

Poster Number: P57
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Clinical Guidelines, Data Sharing, Pediatrics, Clinical Decision Support, Clinical Guidelines, Patient / Person Generated Health Data (Patient Reported Outcomes), Disease Models
Working Group: Clinical Decision Support Working Group
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Background: Congenital diaphragmatic hernia (CDH), a rare anomaly causing partial or complete diaphragm absence, hinders lung development in newborns, with an incidence of 1 in 2,500 to 1 in 3,000 live births. Established in 1995, the Congenital Diaphragmatic Hernia Study Group (CDHSG) collects global data, advocating for multidisciplinary management across 147 centers worldwide. To address data management gaps, we aim to create a scalable, secure, and accessible cloud infrastructure for CDH patient data. This initiative seeks to foster global collaboration, enable predictive modeling for patient outcomes, and advance personalized treatment strategies for CDH. We constructed the CDH registry on a cloud platform providing efficient database management, machine learning, data analysis, encryption, access control, a user-friendly web interface, and an API for data exchange. This design enhances usability and information dissemination within and beyond the CDH community.
By June 2023, the CDH Study Group (CDHSG) had registered over 14,000 patients globally, establishing a leading CDH patient database. Initial analyses have revealed insights into treatment outcomes determinants and opportunities for clinical practice enhancements. The scalable infrastructure supports ongoing data repository expansion and the integration of advanced analytics, including machine learning predictive models for outcome estimation based on patient traits and treatments. The CDHSG consolidates patient data from various sources, including clinical diagnoses, longitudinal studies, registries, and electronic health records. Future plans involve leveraging the database and foster connections with other organizations. Standardizing and amalgamating data aim to develop tools accessible to the wider community, expediting drug development and other advancements.

Speaker(s):
jinlian wang, PhD
UTHealth

Author(s):
Hui Li, Phd - University of Texas Health Science Center at Houston; Matthew Harting, MD, PhD - McGovern Medical School at the University of Texas Health Science Center at Houston; Hongfang Liu, PhD - University of Texas Health Science Center at Houston;

The ED Follow Up Clinic: A Unique Referral Order to Schedule Patients with Follow Up Appointments Prior to Leaving the ED

Poster Number: P58
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Workflow, Clinical Decision Support, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Patient Safety, Transitions of Care, Healthcare Economics/Cost of Care
Primary Track: Applications
Programmatic Theme: Clinical Informatics

In order to reduce the number of return ED visits for lower acuity reasons, we established the ED Follow Up Clinic and created a referral order in our electronic health record (EHR). The referral enabled follow up appointments to be scheduled prior to the patient being discharged from the ED. Over four months, 219 referrals were placed. Only three of these patients were referred back to the ED.

Speaker(s):
Moira Smith
University of Virginia Health System

Author(s):
Sarah Wendel, MD, MBA - University of Virginia Health System; Lauren Smaltz, MD - University of Virginia Health System;

A Conceptual Model of Precision Feedback for Social Determinants of Health Screening and Referral in Primary Care

Poster Number: P59
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Behavioral Change, Knowledge Representation and Information Modeling, Human-computer Interaction
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

We iteratively developed a conceptual model for designing a precision feedback system that generates high-value feedback messages to primary care providers to support individual quality improvement and learning with data about their clinical practice of social determinants of health (SDoH) screening and referral, based on a scoping review of performance feedback about SDoH screening and referral, theories of feedback, motivation and behavior change, and findings from our previous work on precision feedback.

Speaker(s):
Yidan Cao, MS, MPP
University of Michigan Medical School

Author(s):
Yidan Cao, MS, MPP - University of Michigan Medical School; Zach Landis-Lewis, PhD,MLIS - University of Michigan;

LLM-Based Synthetic Tabular Data Generation for Health Equity

Poster Number: P60
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Machine Learning, Fairness and Elimination of Bias, Health Equity, Clinical Decision Support
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

Disparities in model performance for marginalized groups in healthcare prediction tasks are often due to lower representation of these groups in health datasets. We leverage OpenAI’s GPT4 model to generate synthetic data specific to smaller subgroups. Augmenting health datasets with this synthetic data can improve model performance for smaller groups, even compared to other augmentation and reweighting baselines. LLMs hold great promise in synthetic data generation, particularly in sparse-data environments arising from health equity challenges.

Speaker(s):
Daniel Smolyak, M.S.
University of Maryland, College Park

Author(s):
Margret Bjarnadottir, PhD - University of Maryland, College Park; Ritu Agarwal, PhD - Johns Hopkins University; Kenyon Crowley, PhD, MBA, CPHIMS - Accenture Federal Services; Arshana Welivita, PhD - Johns Hopkins University;

Learning a latent confounding representation in high-dimensional observational studies with EHRs via a variational autoencoder

Poster Number: P61
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Deep Learning, Causal Inference
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

Propensity score adjustment is a potential solution to address indirectly measured confounders in observational studies, but may face problems in the setting of high dimensions and small sample size. Here, we focus on dimensionality reduction using a variational autoencoder (VAE) for confounding adjustment as an alternative. We find that VAE-generated latent variables effectively capture information from an indirectly measured confounder. This suggests that there is feature correlation implicit in high-dimensional electronic health records data.

Speaker(s):
Hsin Yi Chen, B.S.
Columbia University

Author(s):
Hsin Yi Chen, B.S. - Columbia University; Linying Zhang, PhD - Washington University in St. Louis; George Hripcsak, MD - Columbia University Irving Medical Center;

ChatGPT for Expansion of Abbreviations and Acronyms in Emergency Department Triage Notes

Poster Number: P62
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Natural Language Processing, Evaluation
Primary Track: Applications
Programmatic Theme: Clinical Informatics

We submitted emergency department triage notes to ChatGPT for expansion of abbreviations and acronyms as a preprocessing step in a larger NLP project. We found that ChatGPT did very well at expanding the abbreviations and acronyms correctly with context awareness. It also corrected spelling errors in the text submitted. However, it introduced some errors and may not be suitable for more nuanced tasks.

Speaker(s):
Phillip Asaro, MD
Washington University in Saint Louis

Author(s):
Albert Lai, PhD, FACMI, FAMIA - Washington University; Kevin Heard, MS HCIN, ACHIP - BJC HealthCare; Aaron Papp - Washington University at St. Louis;

Co-designing Sensorization Plans in the Nursing Home Setting

Poster Number: P63
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Internet of Things, Qualitative Methods, Surveys and Needs Analysis, Nursing Informatics
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Nursing home settings face many challenges such as high staff turnover. Recent healthcare technology advancements present sensorization opportunities for passive data collection to improve patient and staff outcomes, but there is limited empirical evidence on user acceptance. Without formative research introducing health technologies can have unintended consequences. Our rapid ethnographic research is aimed at gathering perspectives from nursing home stakeholders to co-design a sensorization plan to address a priority goal in the community.

Speaker(s):
Lewis Davis, PhD - Human Centered Computing
University of Maryland, Baltimore County

Author(s):
Zainab Balogun, None - University of Maryland Baltimore County; Luke Zimmerman, BA - University of Maryland Baltimore County; Yoon Chung Kim, MHS, MS - University of Maryland; Roberto Yus, PhD - University of Maryland Baltimore County; Kelly Doran, PhD, RN - University of Maryland; Sarah Holmes, PhD, MSW - University of Maryland; Tera Reynolds, PhD, MPH, MS - University of Maryland Baltimore County;

A Framework for genAI-Clinician Collaboration in EHR Systems

Poster Number: P64
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Documentation Burden, Human-computer Interaction, Administrative Systems, Simulation of Complex Systems
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

In recent years, there has been increased interest in the clinician’s role in electronic health record (EHR) systems and their effect on documentation burden, time management, clinician stress, and burnout. Yet, we lack an understanding of how clinicians experience decision-making using EHR systems. The rapid exploration of adopting generative artificial intelligence (genAI) in healthcare leads to consideration of a question overlooked in the early development of EHR systems: “How does documentation change clinician behavior?”

Speaker(s):
Michael Cauley, PhD
Vanderbilt University Medical Center

Author(s):
Richard Boland, PhD - Case Western Reserve University; Corinne Coen, PhD - Case Western Reserve University; Yunmei Wang, PhD - Case Western Reserve University; David Aron, MD, MS - Case Western Reserve University; Dan France, PhD - Vanderbilt University Medical Center; Michael Ward, MD, PhD - Vanderbilt University Medical Center;

Leveraging Graph Structures for Information Retrieval: the TIPTOE Knowledge Path Explorer

Poster Number: P65
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Knowledge Representation and Information Modeling, Controlled Terminologies, Ontologies, and Vocabularies, Healthcare Quality, Surgery, Information Visualization
Primary Track: Applications
Programmatic Theme: Clinical Informatics

The NIH-funded Trauma Institutional Priorities and Teams for Outcome Efficacy (TIPTOE) project (R01GM111324) examines the impact of organizational features in Level 1 and 2 trauma centers on patient outcomes. We hypothesize that variability of organizational features of these trauma centers is a significant factor in the variability of patient outcomes in trauma care. We are recruiting 230 trauma centers for data collection. We aim to provide quick dissemination of actionable information on which organizational factors impact patient outcomes. To do so we have developed the TIPTOE Knowledge Path Explorer KPE). It generates a knowledge graph for each trauma center and allows the trauma stakeholder to explore information in the knowledge graph choosing from multiple starting points, such as the user’s institution or a specific patient outcome measure, including but not limited to mortality, length of stay, acute kidney injury, myocardial infarction, pulmonary embolism, and severe sepsis. Users navigate the data moving from one node to the next using the relations captured by knowledge graph, choosing which edge or relation they want to follow from each node. Using the relations between organizational features and patient outcomes, which have been generated by statistical analysis, users can see how the institution’s organization affects patient outcomes. The KPE provides a visualization of an RDF-based knowledge graph. Thus, this presents a novel technology to make RDF data more accessible and useful for trauma center stakeholders. This technology can be translated to other use cases and areas of patient care.

Speaker(s):
Mathias Brochhausen, Ph.D.
University of Arkansas for Medical Sciences

Author(s):
Justin Whorton, B.S. - University of Arkansas for Medical Sciences; Jonathan Bona, PhD - University of Arkansas for Medical Sciences (UAMS); Reza Shahriari, B.S. - University of Florida; Eric Ragan, Ph.D. - University of Florida; Kevin Sexton, MD - University of Arkansas for Medical Sciences (UAMS); Mathias Brochhausen, Ph.D. - University of Arkansas for Medical Sciences;

Managing Clinical Alerts Feedback: from the Sender to the Changes

Poster Number: P66
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Patient Safety, Workflow, Evaluation, Change Management
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Electronic feedback from users receiving passive and interruptive alerts while using the EHR can be a powerful tool to maintain a large CDS system. However, managing a large number of feedback may be a complex task in need of a managing tool. Here we present a novel application, TREMBUS, designed to facilitate feedback management and actions to improve the quality of the alerts and prevent alert fatigue.

Speaker(s):
Paul Johnsen, BSEE
Mayo Clinic

Author(s):
Paul Johnsen, Mr. - Mayo Clinic; Kathie Schilling, Ms - Mayo Clinic; Carol Eichenlaub, R.N. - Mayo Clinic; Courtney Holbrook, Ms - Mayo Clinic; Pedro Caraballo, MD - Mayo Clinic;

Adapting a scale to assess clinicians’ situational trust in AI

Poster Number: P67
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Human-computer Interaction, Machine Learning
Primary Track: Applications

Successful implementation of artificial intelligence (AI) in healthcare necessitates understanding human-AI interaction. Trust in AI-based clinical decision support is crucial but understudied among clinicians. We will adapt and validate an existing trust scale from the autonomous vehicle domain to assess clinicians’ situational trust in an AI-derived early warning score (EWS). Survey methodology will be employed for scale adaptation and validation and we will share the results of the validated 6-item trust measure.

Speaker(s):
Elizabeth Sloss, PhD, MBA, RN
University of Utah

Author(s):
Usman Sattar, MBBS, MSHI; Guilherme Del Fiol, MD, PhD - University of Utah; Karl Madaras-Kelly, PharmD, MPH - Idaho State University; Jorie Butler, PhD - University of Utah;

Integrating Ontological and Probabilistic Knowledge in Rare Diseases

Poster Number: P68
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Causal Inference, Knowledge Representation and Information Modeling, Controlled Terminologies, Ontologies, and Vocabularies
Primary Track: Applications
Programmatic Theme: Clinical Informatics

We sought to develop a Bayesian-network framework to integrate hierarchical and probabilistic knowledge from ontologies of rare diseases and phenotypic abnormalities. A pilot model incorporated 21 intestinal polyposis syndromes and their 304 associated phenotypic abnormalities with their hierarchical and causal relationships. The model successfully computed disease probabilities based on phenotypic data. Work is underway to extend the framework to more than 7000 rare diseases and associated findings.

Speaker(s):
Charles Kahn, MD, MS, FACMI
University of Pennsylvania

Author(s):
Cheng Thao, PhD - Metro State University; Syed Kamran Khatai, B.Eng. - Metro State University;

Intensive Care Unit (ICU) Readmission or Mortality Risk Model Performance

Poster Number: P69
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Internal Medicine or Medical Subspecialty, Workflow
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Over 5 million Americans annually enter the ICU, with a significant percentage experiencing readmission, associated with higher mortality rates and resource utilization. A retrospective analysis of ICU stays validated a predictive model for readmission or mortality within five days of discharge. Practical implications, such as alert frequency and false positives, highlight the need for careful planning before model implementation to maximize effectiveness and minimize unintended consequences.

Speaker(s):
Vivian Anderson, MD
Wake Forest Baptist Hospital

Author(s):
Vivian Anderson, MD - Wake Forest Baptist Hospital; Lauren Witek, MStat - Atrium Health Wake Forest Baptist; Richa Bundy, MPH - Wake Forest Baptist Health; Tyler George, BS - Wake Forest School of Medicine; Corey Obermiller, MAS Applied Statistics - Wake Forest School of Medicine; Adam Moses, MHA - Wake Forest Baptist Medical Center; Brad Rowland, MD - Atrium Health Wake Forest Baptist; Ajay Dharod, MD - Wake Forest University School of Medicine; Jess Palakshappa, M.D. - Department of Internal Medicine, Section of Pulmonology, Critical Care, Allergy and Immunologic Diseases, Wake Forest School of Medicine;

Disparities in the Documentation of Social Determinants of Health ICD-10 Z-codes for Patients Diagnosed with Cancer: An Epic Cosmos Study

Poster Number: P70
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Health Equity, Racial Disparities, Population Health, Personal Health Informatics, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Data Mining, Data Sharing, Information Visualization
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Understanding social determinants of health (SDOH) is crucial for effective cancer care. To better understand the utilization of SDOH ICD-10 Z-codes, we created a cohort of 182,183,658 individuals who are 18 years or older and have at least one clinical visit between 2014 and 2023 in the multi-institution HIPAA-limited Epic Cosmos database. We found that patients with cancer have disproportionately higher rates of SDOH documentation compared with patients without cancer, especially in problems with (a) Z57-Occupational exposure, (b) Z60-Social environment, (c) Z72-Lifestyle, and (d) Z75-Medical facilities. Our analysis reveals that patients with cancer who identified as Black or African American and American Indian or Alaskan Native have the highest rate of Z-codes documentation (10.2% and 11.3%, respectively), and the two groups also had the highest percentage increase over the 10 years. In addition, individuals who are more socially vulnerable (based on the social vulnerability index) are more often documented with SDOH (17.5% for high risk vs. 10.7% for low risk). We observed an increasing trend of Z-codes documentation among patients with cancer, which underscores the heightened awareness of SDOH as well as the growing utilization of associated Z-codes. It suggests an optimistic future where SDOH factors will be better addressed in cancer care. However, the overall Z-code utilization is still low. Marked social disparities (based on documented SDOH Z-codes) exist for cancer patients and individuals in disadvantaged ethnic and social subgroups. These disparities call for more focused approaches to study root causes and promote equity in cancer care.

Speaker(s):
Yiming Zhang, M.D.
UMass Chan Medical School

Author(s):
Yiming Zhang, M.D. - UMass Chan Medical School; Charlotte DeLeo, MD, MPH - UMass Chan Medical School; Mara Epstein, ScD, ScM - UMass Chan Medical School; Sarah Cutrona, MD, MPH - University of Massachusetts Medical School; Huanmei Wu, FAMIA, PhD - Temple University; Ben Gerber, MD, MPH - University of Massachusetts Chan Medical School; Eric Alper, MD - UMass Memorial Health Care; Feifan Liu, PhD - University of Massachusetts Chan Medical School;

Accuracy of GPT vs Google Translate for Translation of Patient-Specific Emergency Discharge Instructions

Poster Number: P71
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Health Equity, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Delivering Health Information and Knowledge to the Public
Primary Track: Applications
Programmatic Theme: Clinical Informatics

We aimed to compare translation accuracy of Google Translate (GT) versus OpenAI GPT. Using 50 real emergency department patient discharge instructions, we evaluated the accuracy of GT vs GPT for translation into Chinese, Spanish, and Russian. For Chinese translations (all analyses will be complete by summer 2024), GPT was significantly more accurate than GT (94.6% vs 89.5%, p=0.018). Grammar anomalies and sentences providing return precautions were significantly associated with inaccurate translations by GT.

Speaker(s):
Elaine Khoong, MD, MS
University of California San Francisco

Author(s):
Andersen Yang, MPH - UCSF; Jaskaran Bains, MD - UCSF; Ana Milisavljevic, MD - UCSF; Katherine Brooks, MD - UCSF; Marianna Kong, MD - UCSF; Alicia Fernandez, MD - UCSF;

Impact of a Tiered Clinical Decision Support System for Acute Kidney Injury in Cardiac Surgery Patients

Poster Number: P72
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Informatics Implementation, Critical Care, Healthcare Quality
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Nephrotoxic medications may impair renal recovery if continued in acute kidney injury (AKI) patients. This single-center study aimed to improve nephrotoxin discontinuation in cardiac surgery patients with AKI by implementing a tiered clinical decision support system consisting of an AKI alert, nephrotoxin alert, and AKI checklist. Nephrotoxin discontinuation improved over time across Group I [pre-AKI alert (26%)], Group II [post-AKI alert (33%)], Group III [pre-nephrotoxin alert (55%)], and Group IV [post-nephrotoxin alert (67%)](p=0.0018).

Speaker(s):
Christopher Justice, BSN, RN - CRNA/DNP student
WVU Medicine

Author(s):
Connor Nevin, MD Student - West Virginia University; Rebecca Neely, BSN, RN - West Virginia University; Brian Dilcher, MD, FACEP - West Virginia University; Nicole Kovacic Scherrer, PharmD, BCCCP - West Virginia University; Heather Carter-Templeton, PhD, RN, FAAN - West Virginia University; Aaron Ostrowski, DNP, APRN, CRNA - West Virginia University; Gordon Smith, MD, MB, MBChB, MPH - West Virginia University; Jami Pincavitch, MD - West Virginia University; Roopa Kohli-Seth, MD - Icahn School of Medicine at Mount Sinai; Girish Nadkarni, MD, MPH - Icahn School of Medicine at Mount Sinai; Khaled Shawwa, MD, MS - West Virginia University; Ankit Sakhuja, MBBS, MS - Icahn School of Medicine at Mount Sinai;

Developing a Maturity Model for AI Readiness in Healthcare Organizations

Poster Number: P73
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Governance of Artificial Intelligence, Informatics Implementation, Qualitative Methods
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This project aims to develop a maturity model for AI use in healthcare, guiding organizations in assessing their readiness for deploying AI technologies. The methodology used includes literature reviews, semi-structured interviews, and the formation of multi-disciplinary working groups. Preliminary findings of landscape analysis reveals gaps in existing models and help identify five new domains. Next, we aim to refine the model through the collaborative efforts of the workgroups, incorporating the preliminary results.

Speaker(s):
Yufei Long, MPH
Vanderbilt University Medical Center

Author(s):
Laurie Novak, PhD - Vanderbilt University Medical Center Dept of Biomedical Informatics; Matthew Elmore, ThD; Nicoleta Economou-Zavlanos, Ph.D.; Peter Embí, MD - Vanderbilt University Medical Center; Boyd Knosp, MS, FAMIA - University of Iowa; Megan Salwei, PhD - Vanderbilt University Medical Center; Joyce Harris, MA - Vanderbilt University Medical Center;

Exploring Successful Change Management to Implement Artificial Intelligence Technology

Poster Number: P74
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Change Management, Workflow, Qualitative Methods
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Many AI technologies in healthcare have shown promise in research but have failed clinical translation. While 60% of change initiatives fail due to change management, the optimal approach to change management when implementing AI technologies into clinical workflows is unknown. We present findings from a qualitative study within our organization researching successful and unsuccessful approaches when implementing AI tools into clinical workflows across different specialties.

Speaker(s):
Sophie Cameron, MB, ChB, MS
Mayo Clinic

Author(s):
Lauren Rost, PhD - Mayo Clinic; Sophie Cameron, MBChB, MS - Mayo Clinic;

Engaging Cancer Patients in Documentation of Contrast Adverse Events Using PRO-CTCAE Terminology

Poster Number: P75
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Mobile Health, Health Equity, Human-computer Interaction, Patient / Person Generated Health Data (Patient Reported Outcomes)
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Patient history and accurate documentation of prior contrast adverse events are essential to providing quality cancer treatment and safe diagnostic imaging. However, there are current gaps in the documentation of contrast-related adverse events. This project aims to increase patient engagement in the documentation of contrast-related adverse events by building an AI chatbot for data collection using mobile devices. Future goals include the creation of a FHIR resource for EHR data interoperability.

Speaker(s):
Cesar Lam, MD
H. Lee Moffitt Cancer Center & Research Institute

Author(s):
Christina Eldredge, MD, PhD, FAMIA - University of South Florida; James Andrews, PhD, MLIS, FAMIA - University of South Florida, School of Information; Evan Ratkus, Undergraduate - University of South Florida; Jill Jariwala, MSHI - University of South Florida;

Enhancing Disaster Health Data Management using FHIR in Indonesia: Integration of WHO EMT MDS into SATUSEHAT Platform

Poster Number: P76
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Standards, Documentation Burden, Interoperability and Health Information Exchange
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This study integrates the World Health Organization's Emergency Medical Teams Minimum Data Set (WHO EMT MDS) forms into Indonesia's SATUSEHAT platform to standardize health disaster management data collection. By mapping and adjusting WHO and ASEAN EMT MDS variables to FHIR resources, we developed a disaster profile for SATUSEHAT, successfully bridging the gap in data standardization and potentially improving disaster response efficiency.

Speaker(s):
Hiro Faisal, MD
Tohoku University

Author(s):
Masaharu Nakayama, MD, PhD, FAMIA - Tohoku University;

Predicting Mortality for MIMIC Patients with Tuberculosis in ICU using Machine Learning

Poster Number: P77
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Machine Learning, Deep Learning
Primary Track: Foundations

Tuberculosis (TB) is a transmissible disease and ranks among the top ten causes of morbidity and mortality worldwide. The need to strengthen tuberculosis prediction is high in several countries. Accurate prediction of death risk can alleviate situations where patients receive inadequate attention or excessive attention due to their risk level, resulting in long queues for high-risk individuals. To conduct a study on tuberculosis patients, this paper utilized the Medical Information Mart for Intensive Care (MIMIC) database installed by Dr. Joseph Miles from SUNY Oswego. The dataset was obtained from the MIMIC database and various machine learning methods were employed to predict tuberculosis severity in patients admitted to the ICU, including mortality and length of ICU stay. The machine learning models used comprised Linear Regression, Logistic Regression, K-Nearest Neighbor (KNN), Random Forest, Artificial Neural Network (ANN), and Long Short-Term Memory (LSTM). Results showed that LSTM achieved the highest AUC-ROC of 99.1% for classifying mortality among tuberculosis patients. For the regression problem, KNN yielded the best R-squared of 47.4% when predicting the number of days in the ICU.

Speaker(s):
Victoria Nguyen, Data Analyst/Master of Science in Biomedical and Health Informatics
M.S. Hall + Associates, LLC

Author(s):
Victoria Nguyen, Data Analyst/Master of Science in Biomedical and Health Informatics - M.S. Hall + Associates, LLC; Isabelle Bichindaritz, PhD - State University of New York;

Utility of the Fast Healthcare Interoperability Resources (FHIR) Clinical Reasoning Module for Representation of Arden Syntax Medical Logic Modules

Poster Number: P78
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Standards, Interoperability and Health Information Exchange
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Context: Arden Syntax encodes knowledge as Medical Logic Modules, including an uncommon XML alternative format. Objective: Assess whether the FHIR R4 Clinical Reasoning Module can replace this. Methods: Slots of 325 MLMs were extracted and examined. Result: Arden Syntax MLMs can be represented by the PlanDefinition and GuidanceResponse resources with the exception of the Resources category and the validation and type slots. Conclusion: FHIR R4 adequately represents a robust corpus of Arden Syntax MLMs.

Speaker(s):
Robert Jenders, MD, MS, FACP, FACMI, FHL7, FAMIA
Charles Drew University/UCLA

Author(s):
Robert Jenders, MD, MS, FACP, FACMI, FHL7, FAMIA - Charles Drew University/UCLA;

Project Kamogano: Evaluating Healthcare Data Collection, Sharing, and Use in Botswana

Poster Number: P79
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Evaluation, Data Standards, Global Health
Primary Track: Applications
Programmatic Theme: Clinical Informatics

In Botswana, transitioning from paper-based to digital health systems presents opportunities and challenges for healthcare delivery. "Project Kamogano" explored this shift, focusing on data flow and healthcare workers' perspectives. A mixed-methods study involving 150 professionals revealed a positive outlook on digital technology but identified barriers to data sharing, access, and training needs. The findings suggest that targeted educational and infrastructure enhancements are essential for leveraging digital advancements in healthcare.

Speaker(s):
Badisa Mosesane, Bachelors
CHOP

Author(s):
J. Grey Faulkenberry, MD, MPH - Children's Hospital of Philadelphia;

An Episode-Based Analytic Approach to Examine Real-World Cost Impacts of Virtual-First Care for Common Acute Conditions

Poster Number: P80
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Economics/Cost of Care, Telemedicine, Real-World Evidence Generation
Primary Track: Applications
Programmatic Theme: Clinical Informatics

An informatics-informed evaluation leveraging a rich foundation of medical claims and advanced analytics, including episode grouper methodology, was used to determine the real-world cost impact of virtual care as the initial modality of care for treating top common acute conditions compared to in-person-first care among ~240k members of a large national payor residing across the United States.

Speaker(s):
Kelly Craig, PhD
CVS Health

Author(s):
Amanda Zaleski, PhD - Aetna; Xinbei Guan, MS - CVS Health; Kelly Craig, PhD - CVS Health; Christopher Junk, MA - CVS Health; Arthur McGill, MD - CVS Health; Henry Gordon, BS - CVS Health; Dorothea Verbrugge, MD - CVS Health; Kristofer Caya, MBA - CVS Health;

Exploring the Benefits and Risks of Recording Clinical Consultations: A Systematic Review

Poster Number: P81
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Legal, Ethical, Social and Regulatory Issues, Privacy and Security, Patient Engagement and Preferences, Telemedicine, Informatics Implementation, User-centered Design Methods, Clinical Decision Support
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This systematic review assesses the benefits and risks associated with audio or video recording of clinical consultations. Through an analysis of a decade's worth of literature (1956 studies screened to date), it delineates the positive outcomes for patient comprehension and the challenges related to privacy and legal considerations. The findings aim to shape healthcare policy and ethical AI application, enabling stakeholders to effectively balance the delivery of quality healthcare with legal and privacy challenges.

Speaker(s):
Janelle Painter, BN
Sydney Local Health District

Adrian Boscolo, MBBS
NSW HEALTH

Author(s):
Adrian Boscolo, MBBS - NSW HEALTH; Janelle Painter, Bachelor of Nursing - Sydney Local Health District; Angus Ritchie, MBBS - Sydney Local Health District;

Advanced Detection of Nausea/Vomiting and Anxiety in Patients with Cancer

Poster Number: P82
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Natural Language Processing, Deep Learning, Data Mining, Clinical Decision Support, Bioinformatics, Machine Learning, Informatics Implementation, Nursing Informatics
Working Group: Knowledge Discovery and Data Mining Working Group
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

This study leverages large language models (LLMs) to enhance the detection of nausea/vomiting and anxiety in cancer patients. By augmenting the Bio-Clinical BERT model with extensive clinical data and symptom-specific tuning, it introduces an improved method for identifying symptoms from clinical texts. Comparative evaluations reveal its enhanced efficacy over other models, particularly in recognizing physical symptoms. This underscores the utility of LLMs in detecting symptoms and elevating patient care in clinical settings.

Speaker(s):
Nahid Zeinali, PhD student
University of iowa

Author(s):
Alaa Albashayreh, PhD, MSHI, RN - University of Iowa; Weiguo Fan; Stephanie Gilbertson White, PhD, APRN-BC, FAAN - university of Iowa;

Considerations for Identification of Suicidal Ideation and Behavior using the Electronic Health Record

Poster Number: P83
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Information Retrieval, Patient / Person Generated Health Data (Patient Reported Outcomes)
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

Developing approaches for examining health issues within the electronic health record (EHR) requires an understanding of how these approaches operate within different health systems. Figuring out where information on suicidal thoughts and behaviors is documented provides vital information for effective and efficient use of the EHR for suicide research. This pilot project allowed us to navigate data challenges (e.g., data content, format, and quality) and considerations when identifying suicide-related visits within the EHR.

Speaker(s):
Sarah Arias, PhD
Butler Hospital/Brown University

Author(s):

DAWGs - Data Access Working Groups

Poster Number: P84
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Change Management, Curriculum Development, Standards
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

Atrium Health and Advocate health recently combined and have recognized the pivotal role of direct data access in driving advancements within medical informatics and improving patient outcomes. We aim to outline a strategic approach to initiate a Data Access Working Group (DAWG) within a large health system.

Establishing a DAWG involves overcoming organizational challenges, aligning diverse stakeholders, and navigating regulatory complexities. We share insights from our experience in founding a DAWG across a complex health system, emphasizing key considerations, milestones, and lessons learned.

Our poster will show the guiding principles in the DAWG's formation, highlighting the importance of collaboration and fostering a culture of transparency. Key elements of the DAWG include defining governance structures, addressing data security and privacy concerns, and promoting standardization for streamlined data sharing processes.

We will showcase successful strategies for engaging clinicians, researchers, data scientists, and administrators in the DAWG's mission. Additionally, we will discuss the role of innovative technologies and tools in facilitating secure and efficient data access across diverse health system departments.

We hope observers will gain practical insights into the challenges and opportunities associated with launching a DAWG in a large health system, equipping them with actionable knowledge to initiate similar initiatives in their respective organizations. As health systems continue to navigate the evolving landscape of medical informatics, the establishment of robust data access frameworks becomes imperative for fostering collaboration, accelerating research, and ultimately enhancing patient care.

Speaker(s):
Adam Moses, MHA
Wake Forest Baptist Medical Center

Author(s):
Eric Kirkendall, MD, MBI - Wake Forest Baptist School of Medicine/Advocate Health; Whitney Rossman, MS - Atrium Health; Patricia Corn, RN, MSN, CIPP/US/CHC - Atrium Health Wake Forest Baptist; Jason Durham, MA - Atrium Health; Adam Moses, MHA - Wake Forest Baptist Medical Center;

The Influence of Machine Learning-Decision Support Systems on Nurse's Clinical Decision-Making Process

Poster Number: P85
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Nursing Informatics, Clinical Decision Support, Machine Learning
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Nurses' clinical decision-making processes are of critical importance as each clinical decision directly impacts the health, safety, and well-being of patients. Opportunities for further research has been identified within current literature related to the influence of Machine Learning-Decision Support System (ML-DSS) on nurses' clinical decision-making process. This constructivist grounded theory research study explored nurses' perceived influence of ML-DSS on their clinical decision-making process.

Speaker(s):
Ryan Chan, MScN, RN, PhD(c)
Western

Author(s):
Ryan Chan, MScN, RN, PhD(c) - Western University; Richard Booth, PhD, RN - Arthur Labatt Family School of Nursing, Western University; Gillian Strudwick, PhD, RN, FAMIA, FCAN - Centre for Addiction and Mental Health;

Piloting Telehealth Personalized Health Planning in Shared Medical Appointments for Sickle Cell Disease

Poster Number: P86
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Behavioral Change, Self-care/Management/Monitoring, Telemedicine
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Introduction
Sickle cell disease (SCD) is a genetically inherited disorder of the hemoglobin that is associated with severely shortened lifespan. Conscientious self-management is required to improve quality of life, morbidity and mortality, and cost of care. Application of a telehealth Personalized Health Planning in Shared Medical Appointments (PHP-SMA) program could help improve SCD self-management. Personalized Health Planning (PHP) is a model of care that emphasizes patient-centeredness, prevention, and patient engagement. SMAs are an alternative to traditional one-on-one patient visits that can improve patient engagement and health self-management.

Methods
Study objectives were to: 1) adapt the PHP-SMA curriculum for SCD and 2) pilot test a telehealth PHP-SMA for feasibility and acceptability. We assessed 1) number of participants recruited vs. refused to participate, 2) attendance and retention rates, and 3) acceptability of the telehealth PHP-SMA.

Results
After 63 phone calls, 18 individuals were approached to participate. Seven individuals were consented and 1 withdrew prior to the start of the study. Average attendance during the 8-week program was 4.75 attendees weekly (median 4.5, range 4-6). All 6 (100%) participants attended 5 or more sessions; and 5(83%) participants attended 6 or more of the sessions, which has been associated health improvements in other chronic conditions. Focus groups suggested acceptability of the intervention by participants. Participants reported that the virtual format reduced transportation and scheduling barriers to attendance.

Conclusions
The SCD PHP-SMA was feasible and acceptable. Future studies will further explore efficacy of the telehealth PHP-SMA, including ways to implement the program into community settings.

Speaker(s):
Dominique Bulgin, PhD, RN
The University of Tennessee, Knoxville

Author(s):

Using deep learning to study postpartum depression and comorbid anxiety after childbirth

Poster Number: P87
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Deep Learning, Natural Language Processing, Information Extraction, Internal Medicine or Medical Subspecialty, Informatics Implementation, Clinical Decision Support, Population Health
Working Group: Mental Health Informatics Working Group
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Using a novel natural language processing - deep learning method, named PPD-BERT, we analyzed electronics health records of 60,495 women who had a childbirth for comorbid postpartum depression (PPD) and other mental health disorders. PPD-BERT showed an accuracy of 0.93, precision of 0.94, recall of 0.93, and F1-score of 0.93 in detecting PPD from clinical notes. Relative to a control cohort of patients having no PPD, patients with PPD diagnosed with higher anxiety disorders (5.8% vs 57.5%) and PTSD (0.2% vs 5.9%) during 0-24 months postpartum, strongly supporting previous reports.

Speaker(s):
Prakash Adekkanattu, PhD
Weill Cornell Medicine

Author(s):
Prakash Adekkanattu, PhD - Weill Cornell Medicine; Marianne Sharko, MD - Weill Cornell Medicine; Yiye Zhang, PhD - Weill Cornell Medical College; Braja Gopal Patra, PhD - Weill Cornell Medicine; Dept of Population Health Sciences; Div of Health Informatics; Veer Vekaria; Andrea Temkin-Yu, PsyD - Weill Cornell Medicine; Alison Hermann, MD - Weill Cornell Medicine; Jyotishman Pathak, PhD - Weill Cornell Medical College, Cornell University;

Real-world Comparative Effectiveness of Pharmacological Treatments for Opioid Use Disorder

Poster Number: P89
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Real-World Evidence Generation, Causal Inference, Data Mining
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The opioid epidemic remains a critical public health concern in the United States, with Missouri experiencing a significant burden of opioid-related fatalities. Pharmacological interventions, such as methadone and buprenorphine, have demonstrated efficacy in mitigating opioid use disorder (OUD) risk. Nevertheless, their real-world application and effectiveness, particularly in reducing OUD-related emergency department (ED) visits or hospitalizations, remain under-explored. This study evaluates the comparative real-world effectiveness of methadone versus buprenorphine in preventing OUD-associated ED visits or hospitalizations among OUD patients. Utilizing data from the Barnes Jewish HealthCare system, encompassing 14 electronic health record databases, we conducted a retrospective comparative cohort study. The analysis encompassed 5149 patients treated with methadone and 6947 with buprenorphine. Large-scale propensity score with 1:1 matching was used to control for confounding, and causal survival model was used to estimate individual treatment response. Initial findings suggested no statistically significant difference in the effectiveness of methadone compared to buprenorphine across the entire cohort. However, variability in treatment response was observed. This study underscores the absence of a significant difference in reducing OUD-related ED visits or hospitalizations between methadone and buprenorphine but highlights the importance of personalized treatment approaches due to observed heterogeneity in patient responses.

Speaker(s):
Linying Zhang, PhD
Washington University in St. Louis

Author(s):
Ruochong Fan, MA - Washington University in St. Louis; Devin Banks, PhD - Washington University in St. Louis; Wenyu Song, PhD - Brigham and Women's Hospital, Harvard Medical School; Adam Wilcox, PhD - Washington University in St. Louis; Linying Zhang, PhD - Washington University in St. Louis;

An Application that Extracts and Consolidates Data from Clinical Systems to Enhance the Data Manager Experience of Clinical Trial Data Management

Poster Number: P90
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Information Extraction, Information Retrieval, User-centered Design Methods, Human-computer Interaction, Participatory Approach/Science, Informatics Implementation
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Manual abstraction of data from a site’s clinical systems to a biopharmaceutical firm’s electronic data capture (EDC) system is inefficient and error prone. In partnership with data managers (DMs), we used a user-centered design methodology to create a web application, CTDataHub. CTDataHub extracts and consolidates adverse events (AE), concomitant medications (ConMed), and laboratory results data from clinical systems and displays it in a user-friendly view for easy entry into EDCs.

Speaker(s):
Bo Young Kim, Bachelor of Human-Computer Interaction
Memorial Sloan Kettering Cancer Center

Author(s):
Bo Young Kim, Bachelor of Human-Computer Interaction - Memorial Sloan Kettering Cancer Center; Leemor Yuravlivker, BComm - Memorial Sloan Kettering Cancer Center; Michael Buckley, MS, MBA - Memorial Sloan-Kettering Cancer Center; Nancy Bouvier, Bachelor of Science - Memorial Sloan Kettering Cancer Center; Renata Panchal, Master of Science - Memorial Sloan Kettering Cancer Center; Joseph Lengfellner - Memorial Sloan Kettering Cancer Center; Chanda Delgado, Master of Business Administration - Memorial Sloan Kettering Cancer Center; Stephanie Terzulli, PhD - Memorial Sloan Kettering Cancer Center; Paul Sabbatini, MD - Memorial Sloan Kettering Cancer Center;

Inpatient pediatric prescribing errors in relation to prescriber and context factors

Poster Number: P91
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Pediatrics, Patient Safety, Real-World Evidence Generation
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Introduction. This study aimed to identify EHR metadata patterns of pediatric prescribing errors and their associations with prescriber and context factors.
Methods. We identified 5 indicators of intercepted prescribing errors from inpatient pediatric prescriptions from 2019-2022.
Results. We quantified 1,408,939 prescriptions with 22,836 (1.6%) suspected errors. Wrong dose errors associated with shift hours, consecutive days worked, prior order experience, nights, and weekends.
Discussion. Intercepted prescribing errors may be detected reliably using rules-based tools.

Speaker(s):
Daniel Tawfik, MD, MS
Stanford University School of Medicine

Author(s):
Daniel Tawfik, MD, MS - Stanford University School of Medicine; Liem Nguyen, Undergraduate - Stanford University; Dane Jacobson, MD - Stanford University School of Medicine; A J Holmgren, PhD - University of California, San Francisco; Thomas Kannampallil, PhD - Washington University School of Medicine; Stefanie Sebok-Syer, PhD - Stanford University School of Medicine;

Developing Federated Time-to-Event Clinical Scores Using Heterogeneous Real-World Survival Data

Poster Number: P92
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Clinical Decision Support, Knowledge Representation and Information Modeling
Primary Track: Applications

Survival analysis plays a pivotal role in healthcare, informing crucial decisions by predicting the timing of events like disease onset or mortality. While scoring systems are widely used for risk prediction, existing approaches often assume single-source data, limiting collaboration across multiple sites. To address this, we introduce a novel federated survival scoring system framework. Through testing on emergency department data from Singapore and the United States, our approach ensures privacy while significantly enhancing predictive accuracy.

Speaker(s):
Nan Liu, PhD
National University of Singapore

Author(s):
Siqi Li, Bachelor of Science - Duke-NUS Medical School; Yuqing Shang, MS - Duke-NUS Medical School; Ziwen Wang, PhD - Duke-NUS Medical School; Qiming Wu, MS - Duke-NUS Medical School; Chuan Hong, PhD - Duke University; Yilin Ning, PhD; Di Miao, Master - Duke-NUS; Marcus Ong, MPH - Singapore General Hospital; Bibhas Chakraborty, PhD - Duke-NUS Medical School; Nan Liu, PhD - National University of Singapore;

To SNAP or Not to SNAP: Examining Safety Net Antibiotic Prescription Patterns in Pediatric Acute Otitis Media Using a Large Language Model

Poster Number: P93
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Pediatrics, Infectious Diseases and Epidemiology, Large Language Models (LLMs), Natural Language Processing, Healthcare Economics/Cost of Care, Information Extraction, Real-World Evidence Generation
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The primary aim of this study is to describe safety net antibiotic prescriptions (SNAPs) practice patterns and pharmacy dispense data among three clinical settings in children aged 6 months to 18 years, using a retrospective, observational cohort study design. A HIPAA-secure large language model (LLM) was utilized for chart review to determine if a prescription was a SNAP in place of human readers. LLM classifications were matched with pharmacy dispense data to examine antibiotic usage.

Speaker(s):
Jessica Pourian, MD
UCSF

Author(s):
Jessica Pourian, MD - UCSF; Ben Michaels, PharmD - UCSF; Augusto Garcia Agundez Garcia, PhD - UCSF; Valerie Flaherman - UCSF;

An Ontology-Guided Analysis to Unveil Disparities of Social History Documentation in Discharge Notes

Poster Number: P94
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Diversity, Equity, Inclusion, Accessibility, and Health Equity, Racial Disparities, Controlled Terminologies, Ontologies, and Vocabularies, Information Extraction, Natural Language Processing, Knowledge Representation and Information Modeling
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

This study examines significant disparities in the documentation of social history within discharge notes, with a focus on the Social Determinants of Health (SDoH) across different racial groups. Utilizing the MIMIC-IV database, the research analyzes these disparities through the lens of “behavior and lifestyle”, “economic stability”, and “social and community context”. Our findings highlight a notable variation in documentation across races and underscore the importance of improved documentation practices.

Speaker(s):
Zhaoyi Sun, Master of Science
University of Washington

Author(s):
Zhaoyi Sun, Master of Science - University of Washington; Sitong Liu, Master of Science - University of Washington; Wenyu Zeng, PhD - University of Washington; John Gennari, PhD - University of Washington, Dept of Medical Education;

Predicting Readmission Risk in Patients Undergoing Revascularization for Chronic Limb-Threatening Ischemia Using Machine Learning

Poster Number: P95
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Machine Learning, Chronic Care Management
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

This study employs machine learning to predict 90-day readmission and ED visit risks in CLTI patients post-revascularization surgery, using University of Florida Health IDR data from 2015 to 2023. SVM and Regularized Logistic Regression outperformed other models, with significant predictors including physical distance to UF Health, ER visit history, and specific health conditions. These findings underscore the potential of machine learning in improving clinical decision-making, which can be further improved using post-discharge information.

Speaker(s):
Mei Liu, PhD
University of Florida

Author(s):
Qi Xu, Ph.D - University of Florida; Megan Gregory, Ph.D. - University of Florida; Mei Liu, PhD - University of Florida; Samir Shah, MD, MPH - University of Florida; Ho Yin Chan, PhD - University of Florida; Nicole Hammer, B.A. - University of Florida - Department of Health Outcomes and Biomedical Informatics; Todd Manini, PhD - University of Florida; Miad Alfaqih, Phd - University of Florida; Sean Pajak, B.S. - University of Florida; Chase Antilla, B.S. - University of Florida;

Unifier: multi-center unit harmonization and quality control pipeline for continuous variable measurements in the All of Us research program

Poster Number: P96
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Transformation/ETL, Informatics Implementation, Bioinformatics, Usability
Working Group: Clinical Research Informatics Working Group
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

The All of Us (AoU) Research Program combines electronic health record data from partner sites in the Observational Medical Outcomes Partnership common data model. Researchers must address missing units, multiple units, and incorrect labeling of units for each continuous variable measurement in AoU before using data in their analyses. Unifier standardizes multi-center unit harmonization for continuous variable measurements providing researchers with data in a research-ready format, lowering the barrier to conduct biomedical research across populations.

Speaker(s):
James Brogan, MD, MSc
Vanderbilt University Medical Center

Author(s):

Improving usability of the Pediatric Cancer Data Commons (PCDC) Portal

Poster Number: P97
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Usability, Data Sharing, Human-computer Interaction, Information Visualization, Information Retrieval, Pediatrics, Surveys and Needs Analysis
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The University of Chicago Data for the Common Good team conducted a series of quality improvement sessions between April 2023 and August 2023 to improve usability of the Pediatric Cancer Data Commons (PCDC). The PCDC allows researchers to perform cohort discovery on over 40,000 patients with cancer, filtered by demographics, disease type and other factors. Users can request line-level data for further research. Usability challenges and proposed solutions are presented in this poster.

Speaker(s):
Rolando Palacios, MS
University of Chicago

Author(s):
Kirk Wyatt, M.D. - Sanford Health; Samuel Volchenboum, MD,PhD - University of Chicago; Brian Furner, MS - University of Chicago; Ellen Cohen, MPP - University of Chicago; Suzi Birz, MScMI - HiQ Analytics, llc; Luca Graglia, MS - University of Chicago; Kathryn Bouzein, MS - University of Chicago; Spencer Claxton, BS - University of Chicago;

Associations of Symptom-Functioning Clusters and Survival in Colorectal Cancer Patients

Poster Number: P98
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Self-care/Management/Monitoring, Cancer Prevention, Patient / Person Generated Health Data (Patient Reported Outcomes), Machine Learning
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Introduction
There are still limited studies examining symptom patterns and their impact on survival in colorectal cancer (CRC) patients. Consequently, we aim to identify clusters of symptoms among CRC patients and investigate their association with survival outcomes.
Methods
The Colorectal Cancer Cohort Study (2013-2021) at Seoul National University Hospital included 985 CRC patients, with 573 (58.2%) being male and 414 (42.05%) under 60 years old at baseline. Patients provided self-report data on 33 symptoms and functioning, anxiety and depression using the EORTC QLQ-C30, EORTC QLQ-CR29, and Hospital Anxiety and Depression Scale, which were divided into three domains. The K-modes algorithm was used to identify symptom cluster. Kaplan-Meier analysis and Cox proportional hazards regression analysis determined hazard ratios for each cluster.
Results
Three symptom clusters were identified from cluster analysis: Cluster 1 (prevalence 55.1%; All domains being good), cluster 2 (prevalence 24.5%; Low somatic symptoms, low psychosocial symptoms, and low functions), and cluster 3 (prevalence 20.4%; Low somatic symptoms, high psychosocial symptoms, and low functions). Significant differences were observed between clusters by sex, income level, physical activity, total cholesterol, and cancer stage. Survival outcomes also showed significant differences across clusters (log-rank, P = 0.006). Specifically, patients in Cluster 2 showed a 1.64 times higher risk of mortality compared to those in Cluster 1 (adjusted hazard ratio, 1.62 [95% CI, 1.03-2.61]; P = 0.036) in Table 1.
Discussion and Conclusions
Most CRC patients experienced high somatic symptoms, high psychosocial symptoms, and high functions, which were associated with survival.

Speaker(s):
Ji Won Yu, Master Degree
Hallym University

Author(s):
Jae Hyun Park, MD - Department of Surgery, Seoul National University College of Medicine; Ji Won Yu, Master Degree - Hallym University; Hak Jun Kim, B.E. - Department of Artificial Intelligence Convergence, Hallym University, Chuncheon, Republic of Korea; Nan Song, PhD - College of Pharmacy, Chungbuk National University; Min Jung Kim, MD, PhD - Department of Surgery, Seoul National University College of Medicine; Hyun Sun Jun, BSN - Department of Surgery, Seoul National University College of Medicine; Seung Bum Ryoo, MD, PhD - Department of Surgery, Seoul National University College of Medicine; Kyu Joo Park, MD, PhD - Department of Surgery, Seoul National University College of Medicine; Seung Yong Jeong, MD, PhD - Department of Surgery, Seoul National University College of Medicine; Young Ho Yun, MD, PhD - Department of Family Medicine, Seoul National University College of Medicine; Ji Won Park, M.D., Ph.D. - Seoul National University Hospital; Jin Ah Sim, PhD - Hallym University;

OMOPVocabMapper: A Tool for Mapping ICD Codes to OMOP Concepts

Poster Number: P99
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Controlled Terminologies, Ontologies, and Vocabularies, Data Standards, Interoperability and Health Information Exchange
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Observational health research involves defining study cohorts using codes from healthcare terminologies. We created OMOPVocabMapper, a Julia program that maps ICD-9-CM and ICD-10-CM codes to their equivalent OMOP concept IDs, allowing researchers to leverage phenotype-finding resources such as phecodeX in tandem with OMOP CDM databases and tools. OMOPVocabMapper is increasingly being used by our researchers and its functionality will be expanded to accommodate other vocabularies and identify descendant codes.

Speaker(s):
Mounika Thakkallapally, Master of Science
Brown University

Author(s):
Mounika Thakkallapally, Master of Science - Brown University; Jonah Bradenday, Bachelors of Arts - Brown University; Dilum Aluthge, M.D./Ph.D. Student - Brown University; Neil Sarkar, PhD, MLIS - Rhode Island Quality Institute & Brown University; Karen Crowley, PhD - Brown University - Brown Center for Biomedical Informatics; Elizabeth Chen, PhD - Brown University;

Extraction of structured data from unstructured palliative care consult questions using the MedSpacy NLP Library

Poster Number: P100
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Natural Language Processing, Information Extraction, Internal Medicine or Medical Subspecialty
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Accessing detailed structured data from electronic health records (EHRs) for research and quality improvement is a significant challenge for smaller departments in less resourced institutions. As such, extracting maximal information from available data is key. Using Medspacy, an open source clinical natural language processing library for Python, we developed a rule-based algorithm to extract structured patient data from unstructured palliative care consult questions to create more granular departmental data.

Speaker(s):
Kent McCann, MD
Baystate Medical Center

Author(s):
Tovy Kamine, MD, MBA - Baystate Medical Center;

Hospital Performance on Advanced Clinical Decision Support Areas in the Leapfrog Group’s 2022 CPOE Evaluation Tool

Poster Number: P101
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Patient Safety, Evaluation
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The Leapfrog Group's CPOE Evaluation Tool assesses the ability of hospitals' EHR systems to detect common medication errors. The tool assesses basic and advanced clinical decision support (CDS) areas, and provides hospitals with feedback through an overall percentage score of unsafe medication test orders detected. For this study, we report on hospital performance overall, and in advanced CDS categories across hospital demographic details.

The dependent variables in our study were overall and advanced CDS performance. Our independent variables included: location, healthcare system membership, academic medical center designation, hospital bed size, Joint Commission accreditation, EHR vendor, and EHR install date. We ran t-tests and one-way ANOVA to compare performance across hospital demographics, and a multiple linear regression to analyze potential relationships between advanced CDS scores and select hospital demographics.

The mean overall score was 73.8%, and the mean advanced CDS score was 56.7% (N = 1898). System-member hospitals scored higher overall compared to non-system member hospitals (74.5% vs. 69.1%, p < 0.001). Hospitals using “Vendor A” also had higher overall scores than hospitals using other EHR systems (76.0% vs. 71.7%, p <0.001). From the multiple linear regression, system-member hospitals scored 5.6% higher in advanced CDS categories (OR 5.56, 95% CI 3.34, 7.71). Hospitals using Vendor A score 2.3% higher in advanced CDS categories (OR 2.30, 95% CI 0.624, 3.98).

Although hospitals detected most medication test orders in the tool, there is need for improvement in advanced decision support areas, thus future studies about CDS implementation and patient safety should target these advanced CDS areas.

Speaker(s):
Zoe Co, BS
University of Michigan

Author(s):
David Classen, MD - University of Utah School of Medicine; David Classen, MD, MSc - University of Utah; Melissa Danforth, BA - The Leapfrog Group; David Bates, MD - Brigham and Women's Hospital;

Predicting Glaucoma Visual Field Progression Using Functional Principal Component Analysis

Poster Number: P102
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Bioinformatics, Population Health, Advanced Disease
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

In this study of 2,372 glaucoma patients from the Byers Eye Institute at Stanford Medicine, we showcase the utility of functional principal component analysis in modeling glaucoma progression, predicting over 10 years of future visual field exam metrics for patients from one year of their historical visual field data from electronic health records. This longitudinal approach leverages shared information across patients to tailor predictions for individuals’ future curves and yielded a root mean squared error of 2.99 and an R2 value of 0.7467, surpassing prior benchmarks in the literature.

Speaker(s):
Rithvik Donnipadu, Master's Degree
Stanford University

Author(s):
Rithvik Donnipadu, MS - University of San Francisco; Maxim Sivolella, MS - University of San Francisco; Sophia Wang, MD, MS - Stanford University; Cody Carroll, PhD - University of San Francisco;

GeoPacks: Enriching Patient Data with Pre-linked Demographic Data

Poster Number: P103
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Transformation/ETL, Informatics Implementation, Workflow
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

The NC TraCS Institute enhances research data services by introducing pre-linked geographical demographic information, termed GeoPacks, alongside clinical data from the CDW. Integrating demographic and geographic data, including SVI, ACS, ADI, and RUCA codes, with EHRs offers research advantages. The GeoPack process involves ETL operations, harmonization, and data structuring. Results show significant data enrichment with high linkage rates. While beneficial, challenges such as privacy concerns and data accuracy are important considerations. Overall, GeoPacks enhance research informatics at UNC, reducing barriers for data enrichment and exploration.

Speaker(s):
Adam Lee, MBA
University of North Carolina At Chapel Hill

Author(s):

Addressing Missing Labs in Common Data Models

Poster Number: P104
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Transformation/ETL, Informatics Implementation, Controlled Terminologies, Ontologies, and Vocabularies
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

Laboratory results without a LOINC code are not populated into our institution’s common data models (CDMs), leading to ~77 million results unavailable for research studies. We systematically identified the missing lab components, including criteria for component selection, exclusion of components with protected health information, and created a list of 1568 components to map to standard LOINC codes. Mapping these will add ~47 million lab results to CDMs, improve data completeness, and benefit researchers using CDMs.

Speaker(s):
Samyuktha Nandhakumar, MS
The University of North Carolina at Chapel Hill

Author(s):
Marshall Clark; Adam Lee, MBA - University of North Carolina At Chapel Hill;

Mapping Variability in Tobacco-Related Health Factors: Implications for Structured Data Use in EHR

Poster Number: P105
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Natural Language Processing, Information Retrieval, Data Transformation/ETL, Data Mining, Delivering Health Information and Knowledge to the Public
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

This study explored Health Factors (HFs) within the Veterans Health Administration's Informatics and Computing Infrastructure, focusing on tobacco use to improve HF's utility in research through natural language processing. By analyzing 4,571 HF types from over 119 million visits, our research identified significant variations in HF documentation, including assertions and context-specific comments. These findings highlight the importance of aligning HF structures with their intended use, enhancing patient data accuracy and advancing methodologies in health informatics.

Speaker(s):
Patrick Meyers, MD
Vanderbilt University Medical Center

Author(s):

Rapid support and implementation of the Precision Augmented Screening Intervention (PASI) pilot project through the ENTHRALL Platform at the Department of Veterans Affairs (VA)

Poster Number: P106
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: User-centered Design Methods, Clinical Decision Support, Data Transformation/ETL, Informatics Implementation, Information Extraction, Data Sharing, Workflow
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The VA Boston Informatics Group has built the ENTHRALL platform to facilitate the rapid deployment of health research technologies. We present a use-case for this platform, demonstrating its capabilities for a succinct development timeline, which could bring the VA closer to being a Learning Health System. The use-case, built for the lung cancer screening project Precision Augmented Screening Intervention (PASI), is a pilot application designed to help clinicians identify a more targeted population for screening.

Speaker(s):
Hannah Tosi, M.S.
U.S. Department of Veterans Affairs

Author(s):
Hannah Tosi, M.S. - U.S. Department of Veterans Affairs; Chunlei Zheng, PhD - U.S. Department of Veterans Affairs; Amelia Tarren, MS - U.S. Department of Veterans Affairs; Stephen Miller, MPH - U.S. Department of Veterans Affairs; Svitlana Dipietro, BS - U.S. Department of Veterans Affairs; Oleg Soloviev, MS - U.S. Department of Veterans Affairs; June Corrigan, MS - U.S. Department of Veterans Affairs; Kyle McGrath, M.A.; Hormuzd Katki, PhD - National Cancer Institute; Lauren Kearney, MD - VA Boston Healthcare System; Tanner Caverly, MD, MPH - VA Ann Arbor Healthcare System; Nichole Tanner, MD - VA Charleston Healthcare System; Renda Wiener, MD, MPH - VA Boston Healthcare System; Mary Brophy, MD; Nathanael Fillmore, PhD - VA Boston Healthcare System; Nhan Do, MD - VA Boston Healthcare System; Danne Elbers, PhD - VA Boston Healthcare System;

Deciphering Smoke Signals: Advanced Text Analysis and Visualization of Smoking-Related Health Factors

Poster Number: P107
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Natural Language Processing, Data Transformation/ETL, Delivering Health Information and Knowledge to the Public, Data Mining
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

This study analyzed semi-structured patient data on tobacco use within the Veteran Affairs Informatics and Computing Infrastructure. Leveraging a cohort of over 16 million veterans, advanced text analysis and a novel probabilistic algorithm were utilized to correct inconsistencies, misspellings, and abbreviations in Health Factors (HFs) Content Text. The approach successfully enhanced data normalization and categorization, enabling more accurate risk factor analysis and establishing a pipeline for consistent data extraction and standardization in clinical research settings.

Speaker(s):
Patrick Meyers, MD
Vanderbilt University Medical Center

Author(s):

Variation in Provider Specialization among 139 non-affiliated Real-World Data Contributors

Poster Number: P108
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Reproducibility, Data Mining, Knowledge Representation and Information Modeling
Working Group: Clinical Research Informatics Working Group
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

Oracle EHR Real-World Data (OERWD) is a national de-identified data set useful for observational findings across 139 non-affiliated healthcare systems. Data in OERWD is associated with a de-identified provider ID categorized and documented using National Uniform Claim Committee (NUCC) taxonomy codes. There are 2,421,948 providers in OERWD associated with both their specialty and contributing organization. We profile the data for overall distribution and co-occurrence of specialties, enabling research related to treatment patterns and outcomes.

Speaker(s):
Joshua Koni, MS
Children's Mercy

Author(s):
Joshua Koni, MS - Children's Mercy Hospital; Rose Reynolds, Ph.D. - Children's Mercy Hospital; Janelle Noel-MacDonnell, PhD; Mark Hoffman, PhD - Children's Mercy Kansas City;

Evaluating Consistency in Generative AI-Driven Medication Side Effect Detection: A Comparative Analysis of Gemini Pro and MedLM-large

Poster Number: P109
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Pediatrics, Clinical Guidelines
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

This study presents a comparative analysis of two generative AI models, namely Gemini Pro and MedLM-large, for assessing the presence of side effects documented by pediatricians treating children with ADHD. These models achieved consistent results in identifying the presence of decreased appetite or weight loss, which were present in 31-32% of notes. With further refinement, such approaches can used to detect and predict real-world side effect patterns, enabling personalized medication management and improved patient outcomes.

Speaker(s):
Fatma Gunturkun, PhD
Stanford University

Author(s):

Leveraging LLMs for rapid cross-lingual transfer learning in on-premise clinical NLP models

Poster Number: P110
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Natural Language Processing, Real-World Evidence Generation, Workflow
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

While multilingual Transformer-based NLP models have significantly improved in recent years, they still require annotated data from the target language to reach optimal performance. We present a novel approach for automatically generating labels for documents in the target language by combining a task-specific pretrained Transformer model with a general-purpose large language model. In a proof-of-concept study to determine three diagnosis status attributes in Portuguese, we achieved accuracies of over 95% for each task.

Speaker(s):
John Doole
TriNetX, LLC

Author(s):
David Huebner, PhD - Averbis; Peter Kluegl, PhD - Averbis; Kris Collins, BA - Averbis; Ahson Saiyed, MS - TriNetX; Matvey Palchuk, MD, MS, FAMIA - TriNetX, LLC; John Doole - TriNetX, LLC;

Introducing a Clinical Data Standards Consultation Service to Support Translational Research and Science

Poster Number: P111
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Standards, Real-World Evidence Generation, Data Sharing, Controlled Terminologies, Ontologies, and Vocabularies
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

Data standards are a critical part of research infrastructure and can increase the efficiency and generalizability of research, but are complex and interrelated. Researchers need guidance to identify and apply relevant data standards for their planned research. We implemented a consultation service to support this need and describe our early experience in this poster.

Speaker(s):
Rachel Richesson, PhD, MPH, FACMI
University of Michigan Medical School

Author(s):
David Hanauer, MD - University of Michigan;

Reducing Demographic Bias in De-identification Using Transformer Models

Poster Number: P112
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Natural Language Processing, Machine Learning, Information Extraction, Deep Learning, Large Language Models (LLMs)
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The issue of bias in clinical natural language processing is receiving increasing attention. This study aims to further reduce demographic bias by using transformer-based models. We applied DeBERTa and our models outperformed the results reported in the previous study using static word embeddings. Our experiment results showed that biases across all demographic dimensions (gender, race, popularity, and decade) were reduced.

Speaker(s):
Youngjun Kim, Ph.D
City of Hope

Author(s):
Elizabeth Curtis, MA - The Ohio State University; Jennifer Garvin, PhD, MBA, MA - OSU/Department of Veteran's Affairs;

Unlocking Efficiency: Exploring GPT for Rapid Scientific Paper Screening

Poster Number: P113
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Information Extraction, Large Language Models (LLMs), Workflow
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

In addressing the daunting task of navigating the vast landscape of published studies, we investigated the viability of leveraging GPT-4 to distill crucial insights from abstracts. Testing with 78 abstracts revealed an 88.16% accuracy in information extraction through zero-shot runs, increasing to 91.67% with prompt learning. Despite its limited scale, our study underscores GPT's potential in streamlining the review and screening of scientific literature, offering a promising avenue for future research and application.

Speaker(s):
Hyeoneui Kim, PhD
Seoul National University

Author(s):
Hyorim Yu, NA - Seoul National University; Sunghoon Kang, BS - Seoul National University; Hyeoneui Kim, PhD - Seoul National University;

Machine Learning Methods for Estimating Gestational Age at Birth from Electronic Health Records

Poster Number: P114
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Precision Medicine, Patient Safety
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

Accurate estimation of gestational age at birth is essential to effectively conduct pharmacoepidemiologic studies on drug safety during pregnancy. In this study, we propose a machine learning framework for the estimation of gestational age at birth by leveraging demographic and clinical information associated with mother-child dyads in a large database of electronic health records.

Speaker(s):
Cosmin Bejan, PhD
Vanderbilt University Medical Center

Author(s):
Cosmin Bejan, PhD - Vanderbilt University Medical Center; Amelie Pham, MD - Vanderbilt University Medical Center; Leena Choi, Ph.D. - Vanderbilt University School of Medicine; Sarah Osmundson, MD - Vanderbilt University Medical Center; S. Trent Rosenbloom, MD, MPH, FACMI, FAMIA - Vanderbilt University Medical Center Dept of Biomedical Informatics; Elizabeth Phillips, MD - Vanderbilt University Medical Center;

Assessing the Quality of Synthetic Data Produced by LLaMA and ChatGPT in the Context of Obstetric Stigmatizing Language

Poster Number: P115
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Natural Language Processing, Nursing Informatics
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

This exploratory study assessed the use of LLaMA and ChatGPT to generate synthetic data containing stigmatizing language in obstetrics care. Using three prompting approaches, six sets of synthetic sentences were generated and evaluated by nurse researchers for how similar they are to clinicians’ notes and how stigmatizing the language is. LLaMA's few-shot approach yielded the highest similarity and stigmatizing ratings, significantly outperforming other approaches. ChatGPT showed varied results with no significant differences.

Speaker(s):
Jihye Scroggins, PhD
Columbia University School of Nursing

Author(s):
Jihye Scroggins, PhD - Columbia University School of Nursing; Veronica Barcelona, PhD - Columbia University School of Nursing; Ismael Hulchafo, MS - Columbia University School of Nursing; Sarah Harkins, BSN, RN - Columbia University School of Nursing; Danielle Scharp, MSN, BSN - Danielle Scharp; Hans Moen, PhD - Aalto University; Anahita Davoudi; Max Topaz, PhD, RN, MA, FAAN, FIAHSI, FACMI - Columbia University School of Nursing;

Perspectives on evolving data sharing practices in research: Participants Voice Concerns Around Increased Data Sharing with External Entities

Poster Number: P116
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Patient Engagement and Preferences, Racial Disparities, Legal, Ethical, Social and Regulatory Issues, Surveys and Needs Analysis, Health Equity
Primary Track: Policy
Programmatic Theme: Clinical Research Informatics

This study evaluated public perspectives about evolving data sharing policies that encourage increased data sharing practices in research studies. The updated National Institutes of Health Data Management and Sharing Policy that was released in January 2023 was used as an example of current policies that require funded researchers to make their data available in public repositories. There is a gap in understanding participants' perspectives on the updated research practices and requires further evaluation to ensure the absence of unintended consequences for the participation in research. Our study evaluated participants' preferences for sharing specific types of data with specific groups, and strategies to enhance trust in data sharing practices through an online survey with 610 participants. Our findings highlight notable racial disparities in willingness to share research data with external entities, especially health policy and public health organizations. Black participants were significantly less likely to share most health data with public health organizations, including mental health (OR: 0.543, 95% CI: 0.323–0.895) and sexual health/fertility information (OR: 0.404, 95% CI: 0.228–0.691) compared with White participants. Moreover, 63% of participants expressed that their trust in researchers would improve with the option to select data recipients. Our findings highlight that participants exhibit reluctance to share specific research data types, emphasizing strong preferences regarding external data access. This highlights the need for a critical reassessment of current data sharing protocols to align with participant concerns.

Speaker(s):
Stephanie Nino de Rivera, BA
Columbia University

Author(s):
Yihong Zhao, PhD, MPhil - Columbia University School of Nursing; Natalie Benda, PhD - Columbia University School of Nursing; Sarah Eslami, BS - Columbia University School of Nursing; Meghan Reading Turchioe, PhD, MPH, RN - Columbia University School of Nursing; Ruth Masterson Creber, PhD, MSc, RN - Columbia University;

A Knowledge Graph Driven Approach to Extend BioNLP Annotations to Facilitate the Generation of Clinical Code Sets

Poster Number: P117
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Natural Language Processing, Information Extraction, Controlled Terminologies, Ontologies, and Vocabularies, Knowledge Representation and Information Modeling
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The extraction of clinical entities and code sets from textual data are fundamental in biomedical research utilizing real-world data sources. However, the generation of clinically valid code sets from unstructured text remains a challenge in Biomedical Natural Language Processing (BioNLP). Current BioNLP methods rely on single controlled terminological resources for entity linking, which limits the breadth of code set generation and increases the likelihood of false positives. To address this limitation, we propose a novel approach utilizing a large-scale Knowledge Graph (KG) within a Knowledge Management System to extend BioNLP annotations to generate semantically meaningful code sets. We implemented and applied semantic-based queries designed to systematically traverse the KG across multiple semantic relationships, thereby identifying a comprehensive range of clinical entities and relevant codes from multiple terminologies. Pilot evaluations on disease entity annotations from clinical trial eligibility criteria demonstrated the generation of code sets with an average of 58 codes per set. Evaluation against curated code sets from the Value Set Authority Center releveled a moderate overlap. The findings suggest that leveraging KGs can facilitate the generation of clinically relevant code sets in a semi-automated manner.

Speaker(s):
Ali Daowd, MD, PhD
Semedy, Inc.

Author(s):
Marcelo Fiszman, MD, Ph.D. - Semedy Inc; Charles Lagor, MD, PhD, MBA - Semedy; Saverio Maviglia, MD - Semedy, Inc.; Roberto Rocha, MD, PhD, FACMI - Semedy, Inc.;

Large Language Models Enhance the Identification of Emergency Department Visits for Symptomatic Kidney Stones

Poster Number: P118
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Natural Language Processing, Precision Medicine
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

We evaluate the use of large language models such as ChatGPT, GPT-4, and Llama-2 as well as other machine learning methods to enhance the identification of healthcare encounters for symptomatic kidney stone disease in the acute care setting. This application is essential to effectively conduct studies investigating kidney stone disease risk factors and disease associations.

Speaker(s):
Cosmin Bejan, PhD
Vanderbilt University Medical Center

Author(s):
Cosmin Bejan, PhD - Vanderbilt University Medical Center; Amy Reed, MD - Vanderbilt University Medical Center; Natalie Pace, MD - Vanderbilt University Medical Center; Siwei Zhang, MS - Vanderbilt University Medical Center; Yaomin Xu - Vanderbilt University School of Medicine; Daniel Fabbri, PhD - Vanderbilt University; Ryan Hsi, MD - Vanderbilt University Medical Center;

Using Implementation Science to Design Strategies for Embedding a Safety Planning Digital Intervention into a Psychiatric Emergency Department

Poster Number: P119
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Mobile Health, Informatics Implementation, Qualitative Methods, Behavioral Change
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

Suicide is currently the second leading cause of death among youth and young adults in Canada. At our organization, we co-designed an app to facilitate suicide safety planning intervention. A systematic effort is needed to embed this app into the routine clinical flow. The proposed research aims to leverage implementation science frameworks and collaboration with patients, clinicians and families to integrate the app into routine practice in the local emergency department with digital equity considerations.

Speaker(s):
Hwayeon Danielle Shin, RN MScN PhD(c)
Centre for Addiction and Mental Health

Author(s):
Juveria Zaheer, MD - Institute for Mental Health Policy Research, Gerald Sheff and Shanitha Kachan Emergency Department Centre for Addiction and Mental Health/ Department of Psychiatry, University of Toronto; John Torous, MD - Harvard Medical School; Gillian Strudwick, RN, PhD - Centre for Addiction and Mental Health;

Digital Literacy of Persons with Dementia (PwD): Implications for the Design of Informatics Tools

Poster Number: P120
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Aging in Place, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Transitions of Care, Usability, Human-computer Interaction
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

Dementia is a public health challenge, and informatics tools offer potential to support Persons with Dementia (PwD). Yet, these are not always designed for diverse populations, particularly those who are not familiar with technology, potentially excluding end users with limited digital literacy. We explored definitions, measurement tools, and interventions related to digital literacy. We provided design recommendations for informatics systems to be inclusive for PwD and their families by principles of acceptability, usability, and adaptability.

Speaker(s):
Hannah Cho, MSN
University of Pennsylvania

Author(s):
Emma Cho, PhD, RN - University of Pennsylvania/ School of Nursing; Sang Bin You, MSN, RN - University of Pennsylvania; George Demiris, PhD - University of Pennsylvania;

Who is Willing to Engage with Texting for Diabetes Management: Exploring Intervention Reach

Poster Number: P121
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Self-care/Management/Monitoring, Chronic Care Management, Mobile Health, Education and Training, Health Equity, Informatics Implementation
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

Technology-based interventions can support diabetes self-management but may also exacerbate disparities in accessing care. We must understand if these interventions can reach patients who may benefit the most from them. Being female, younger, and in relatively better diabetes control is associated with increased chance of enrolling in a diabetes texting intervention. Racial/ethnic minorities, income, and rurality were not associated with enrollment, suggesting that texting interventions can reach patients who face unique barriers to diabetes self-management.

Speaker(s):
Stephanie Robinson, PhD
VA Bedford Healthcare System

Author(s):
Ndindam Ndiwane, MPH - VA Bedford Healthcare System; Mark Zocchi, PhD - Veterans Health Administration; Courtney Bilodeau, MPH, RDN - VA Bedford Healthcare System; Zhiping Huo, MS - Edward Hines, Jr. VA Hospital; Linda Am, MPH - VA Bedford Healthcare System; Stephanie Shimada, PhD - Department of Veterans Affairs;

Examining the Equitable Adoption and Utilization of Remote Health Monitoring Systems for Cancer Pain Management

Poster Number: P122
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Real-World Evidence Generation, Mobile Health, Tracking and Self-management Systems, Telemedicine, Advanced Disease
Primary Track: Applications

We deployed a remote health monitoring system, designed to help manage cancer pain, to dyads of patients with advanced cancer and their family caregiver home for two weeks. We tracked cancer pain recorded by patient and caregiver using on-demand ecological momentary assessments. Differences in recorded pain and system utilization highlight the need to tailor the system to user characteristics. Future analysis will further examine the relationship between user characteristics, system utilization, and cancer pain experience.

Speaker(s):
Mina Ostovari, PhD
University of Virginia

Author(s):

Utilizing Fitness Trackers for Early Detection of Mild Cognitive Impairment: A Pilot Study on Non-Invasive Digital Biomarkers

Poster Number: P123
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Mobile Health, Machine Learning, Disease Models, Biomarkers
Primary Track: Applications

Background: Early signs of Alzheimer's disease (AD) are difficult to detect, causing diagnoses to be significantly delayed to time points when brain damage has already occurred, and current experimental treatments have little effect on slowing disease progression. Tracking cognitive decline at early stages is critical for patients to make lifestyle changes and consider new and experimental therapies. Frequently studied biomarkers are invasive and costly and are limited in predicting conversion from normal to mild cognitive impairment (MCI).
Objective: Use data collected from fitness trackers to predict MCI status.
Methods: In this pilot study, fitness trackers were worn by 20 participants: twelve MCI and eight age-matched controls. We collected physical activity, heart rate, and sleep data from each participant for up to a month and further developed a machine learning model to predict MCI status.
Results: Our machine learning model was able to perfectly separate between MCI and controls (AUC=1.0). The top predictive features from the model include peak, cardio and fat burn heart rate zones, resting heart rate, average deep sleep time, and total light activity time.
Conclusions: Our results suggest that a longitudinal digital biomarker differentiates between control and MCI patients in a very cost-effective and noninvasive way and, hence, may be very useful for identifying patients with very early AD who can benefit from clinical trials and new, disease-modifying therapies.

Speaker(s):
Qidi Xu, PhD student
University of Texas Health Science Center at Houston

Author(s):

Digital Inclusion Screening and Competency in Digital Health Management

Poster Number: P124
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Personal Health Informatics, Surveys and Needs Analysis, Patient Engagement and Preferences
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

We administered surveys and task assessments to measure digital inclusion and competency in using digital health management tools among older adults. While the majority were equipped to access digital health management tools, a substantial proportion reported challenges and required tailored or comprehensive assistance to use them. This study will identify digital inclusion measures that can reliably predict competency in using digital health management tools to guide implementation of supportive interventions within health care systems.

Speaker(s):
Lina Tieu, PhD
UC Davis Center for Healthcare Policy and Research

Author(s):
Lina Tieu, PhD - UC Davis Center for Healthcare Policy and Research; Courtney Lyles, PhD - UC Davis, Department of Public Health Sciences / Center for Healthcare Policy and Research; Hyunjin Cindy Kim, MPH - University of California San Francisco; Jeanette Wong, BS - University of California San Francisco; Isabel Luna, BA - University of California San Francisco; Andersen Yang, MPH - University of California San Francisco; Jorge Rodriguez, MD - Brigham and Women's Hospital/Harvard Medical School; Elaine Khoong, MD, MS - University of California San Francisco;

mHealth app-based positive psychology intervention in college students may enhance human flourishing

Poster Number: P125
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Mobile Health, Behavioral Change, Patient / Person Generated Health Data (Patient Reported Outcomes), Personal Health Informatics
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

This study investigated the impact of the Roadmap mobile health app on the mental health of nearly 2,000 college students. The app offers positive psychology activities and integrates with wearable sensors for data collection. Among participants, consistent app users showed increased positive mood and reduced anxiety, correlating with flourishing. Though preliminary, results indicate that mHealth interventions could significantly support college students' mental well-being.

Speaker(s):
Shira Hanauer
Greenhills School

Author(s):

Determinants of Trust: A Large-Scale Quantitative Inquiry into Healthcare Experiences and Patient Trust

Poster Number: P126
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Surveys and Needs Analysis, Data Mining
Working Group: Consumer Health Informatics Working Group
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

Trust in US healthcare has been declining despite significant efforts in patient-centered reforms. To bridge the gap, the current study aimed to identify patient trust determinants through a large-scale analysis with schematized survey data from 5,665 hospitals (N=665,053) using linear machine learning models. Our findings indicate staff collaboration as the primary experiential determinant of trust across care settings and highlight the importance of extensive, quantitative research investigating the relationships between patient-perceived experiences and trust.

Speaker(s):
Sin-Ying (Alina) Lin, PhD
Qualtrics

Author(s):

Personalization in Digital Hypertension Management: Systematic Review

Poster Number: P127
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Precision Medicine, Mobile Health, Chronic Care Management, Self-care/Management/Monitoring, Human-computer Interaction, Patient Engagement and Preferences, Health Equity, Telemedicine
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

This systematic review analyzed personalization characteristics of 20 digital hypertension interventions using the Fan and Poole classification. Findings reveal a mix of delivery methods and personalization strategies, predominantly using behavioral, clinical, and psychological data. The importance of diverse data utilization for improved efficacy is highlighted, alongside under explored areas of environmental and digital literacy data. This underscores the potential for more comprehensive personalization approaches in future digital health interventions.

Speaker(s):
Namuun Clifford, MSN, FNP-C
The University of Texas at Austin

Author(s):

How Quickly Do Cancer Patients View the Results of Their Imaging Studies?

Poster Number: P128
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Patient Engagement and Preferences, Evaluation, Data Mining
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

It is not known how rapidly patients look at results in the portal. At a cancer hospital, we measured the time from when a radiology study was performed until the patient reviewed the result online. We found that 31.1% were read before 8 a.m. the next day, 20.3% were read between 8 a.m. and 5 p.m. the next day, 23.2% were read subsequently, and 25.4% were never read.

Speaker(s):
Gilad Kuperman, MD, PhD
Memorial Sloan Kettering Cancer Center

Author(s):
Gilad Kuperman, MD, PhD - Memorial Sloan Kettering Cancer Center; Allison Lipitz-Snyderman, PhD - Memorial Sloan Kettering Cancer Center; SUSAN CHIMONAS, PhD - Memorial Sloan Kettering Cancer Center; Fernanda Polubriaginof, MD PhD - Memorial Sloan Kettering Cancer Center;

Identifying Adaptations Required to Scale the Role of Digital Navigators in Mental Health Settings

Poster Number: P129
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Informatics Implementation, Mobile Health, Behavioral Change, Real-World Evidence Generation
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

The role of digital navigators (DNs) has recently emerged to support clinicians and patients in utilizing digital technologies in healthcare. Limited literature exists on DNs in health settings, especially in the Canadian context. Given this, a multi-site comparison of a mobile app implementation, led by DNs, was conducted. This poster demonstrates adaptations made to the DN role in each implementation context, where adaptations were made to accommodate differences in human/financial resources, culture, and setting.

Speaker(s):
Gillian Strudwick, RN, PhD
Centre for Addiction and Mental Health

Author(s):
Danielle Shin, RN MScN - CAMH; Wenjia Zhou, MHI - Riverside University Health System; Suzanna Juarez-Williamson, n/a - Riverside University Health System; Navi Boparai, MHA - Centre for Addiction and Mental Health; Gillian Strudwick, RN, PhD - Centre for Addiction and Mental Health; Sean Kidd, PhD - Centre for Addiction and Mental Health;

Evaluating the Role of ChatGPT in Enhancing Alzheimer's Disease Management: A Comparative Study

Poster Number: P130
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Evaluation, Patient Engagement and Preferences
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

This study assessed the accuracy and validity of ChatGPT in managing Alzheimer's disease (AD) by comparing its responses with those from neurologists. Evaluators, including neurologists and patients' families, rated both ChatGPT and neurologists' responses highly. However, ChatGPT received significantly higher ratings (4.4 ± 0.6) compared to the neurologists (3.9 ± 0.7, p<.001) on a 5-point scale. These results suggest that ChatGPT can effectively provide information about AD to patients, their families, and healthcare providers.

Speaker(s):
Jialin Liu, MD
West China Hospital Sichuan University

Author(s):

Will Patients Accept Generative AI Genetic Counseling?

Poster Number: P131
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Delivering Health Information and Knowledge to the Public, Patient Engagement and Preferences, Qualitative Methods
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

This study explores acceptance of AI-generated answers to genetic counseling questions. Patients evaluated human and LLM-generated responses to questions regarding Alzheimer’s risk. LLM responses were rated more relevant, trustworthy, and coherent than those from counselors. While patients seem receptive to LLM-provided information, their preferences for communication style vary, highlighting the need to tailor AI responses. This underscores the potential of LLMs in genetic counseling, warranting further research to prevent inaccuracies and address individual patient needs.

Speaker(s):
Yidi Huang, MS
University of Pennsylvania

Author(s):
Hita Kambhamettu, BS - University of Pennsylvania; Angela Bradbury, MD - Hospital of the University of Pennsylvania; Kevin Johnson, MD, MS - University of Pennsylvania;

Comparison of Cancer Survivor Information Non-Seeking and Seeking: Analysis of HINTS-SEER Data

Poster Number: P132
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Patient Engagement and Preferences, Surveys and Needs Analysis, Cancer Prevention
Working Group: Consumer Health Informatics Working Group
Primary Track: Foundations
Programmatic Theme: Consumer Health Informatics

Cancer information non-seekers are studied less than active information seekers. This study provides updated results on cancer information non-seekers, analzying 1,213 responses from a 2021 dataset that combined HINTS and SEER data, using the Comprehensive Model of Information Seeking (CMIS) as a framework.
We found that 22.2% of cancer survivors were non-seekers of information, down 10% from the previous study conducted two decades ago. Our findings can guide development of interventions tailored to information non-seekers.

Speaker(s):
James Andrews, PhD, MLIS, FAMIA
University of South Florida, School of Information

Author(s):
JungWon Yoon, Ph.D. - Jeonbuk National University, Jeonju-si, South Korea;

Discovery of Bone Marrow Transplant Educational Topic Timing Preferences

Poster Number: P133
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Patient Engagement and Preferences, Education and Training, Informatics Implementation
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

Bone marrow transplant (BMT) centers develop caregiver educational materials, but caregivers report unfulfilled needs and feeling overwhelmed with topic timing. We hypothesize that the transplant community can articulate their preference for when specific topics should be presented. Participants were recruited via email to participate in an anonymous survey on the timing of BMT education topics. Responses provide evidence that standardized timing will benefit some but not all caregivers.

Speaker(s):
John Huber, MS
University of Cincinnati Department of Biomedical Informatics

Author(s):
John Huber, MS - University of Cincinnati Department of Biomedical Informatics; Christopher Dandoy, MD, MS - Cincinnati Children’s Hospital Medical Center; Lisa Vaughn, PhD; Jacob Wall, RN, BSN - Cincinnati Children’s Hospital Medical Center; Gabby OConnor, BS - Cincinnati Children’s Hospital Medical Center; Priscila Badia, MD - Cincinnati Children’s Hospital Medical Center; Judith Dexheimer, PhD - Cincinnati Children's Hospital;

Evaluation of the Quality of Online Health Information for Domestic Violence Circulated in Online Health Communities

Poster Number: P134
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Internet of Things, Delivering Health Information and Knowledge to the Public, Healthcare Quality, Personal Health Informatics, Evaluation
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

This study explores the quality (i.e., credibility and usability) of domestic violence websites shared in online health communities. Most websites met higher quality standards on parameters of usability, yet quality on parameters of credibility varied. The findings highlight the potential implications of online website resources evaluation as there are no clear guidelines on quality for OHCs.

Speaker(s):
Rosanna Tsang, Bachelor of Science in Nursing
Hong Kong Polytechnic University

Author(s):
Rosanna Tsang, Bachelor of Science in Nursing - Hong Kong Polytechnic University; Malavika Eby, Bachelor's degree - Swarthmore College; Bohan Zhang, MM - The Hong Kong Polytechnic University; Rose Constantino, Health & Community Systems, School of Nursing - University of Pittsburgh; Vivian Hui, RN, PhD - The Hong Kong Polytechnic University;

Human Papillomavirus Information Needs Among Reddit Users

Poster Number: P135
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Cancer Prevention, Behavioral Change, Health Equity, Infectious Diseases and Epidemiology, Real-World Evidence Generation
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

This study investigates the information needs regarding HPV among online users, focusing on Reddit. Data were collected from postings between November 2021 and January 2024, with 'HPV' used as a keyword for screening. Thematic analysis was conducted to examine the data. From 363 quotes, 4 themes emerged. This study highlights substantial information needs regarding HPV. Healthcare providers must address through education when recommending the vaccine and the HPV exam.

Speaker(s):
Soojung Jo, PhD RN
Purdue University

Soojung Jo, PhD RN
Purdue University

Author(s):
Min Sook Park, PhD - Florida State University, School of Information;

Exploring Health Consumers’ HPV Vaccine Awareness from Reddit

Poster Number: P136
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Cancer Prevention, Data Mining, Machine Learning, Real-World Evidence Generation, Health Equity, Infectious Diseases and Epidemiology
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

The purpose of this study is to better understand lay health consumers’ awareness of HPV vaccines focusing on risks and benefits. Employing text mining techniques, associations among terms with those terms that represent health consumers’ awareness of the HPV vaccines were explored. Overall, awareness of the purpose of the HPV vaccination was low.

Speaker(s):
Soojung Jo, PhD RN
Purdue University

Soojung Jo, PhD RN
Purdue University

Author(s):
Min Sook Park, PhD - Florida State University, School of Information; Christopher Farber, HS - Purdue University; Neelesh Sarathy, HS - Purdue University; Selina Lin, HS - Purdue University;

Role of Digital Health Literacy in Online Health Information for Korean American Older Adults

Poster Number: P137
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Health Equity, Aging in Place, Patient Safety
Working Group: Nursing Informatics Working Group
Primary Track: Foundations

Aim: Lack of digital health literacy among KA older adults poses a significant barrier to their information and technologies. Digital health literacy is essential for older adults to effectively use online information in the context of healthcare. The purpose of this study is to examine technology use, digital health literacy, and experiences with using technology for online health information among Korean American (KAs) older adults. Methods: The study employed surveys to gather data from KAs aged 50 or above. In total, 99 participants successfully completed the survey. Results: The results of the study revealed that the KAs older adults expressed confidence in their ability to find and use online health information, as well as their knowledge of available health resources on the internet. However, they felt that they lacked skills in evaluating the quality of online health resources and distinguishing between high and low-quality sources. Additionally, they lacked confidence in using information from the internet to make health decisions. Additionally, they may require assistance in translating the information they find into actionable health decisions. Digital health literacy varied among different demographic groups within the KA older adult population, with education level playing a significant role. Conclusion: This study emphasizes the significance of understanding and addressing the digital health literacy requirements of older adults in the KA community. By gaining insights into their digital health literacy skills, interventions can be developed to empower this population to effectively utilize health information and make informed decisions about their health.

Speaker(s):
HyeJin Park, PhD RN
FSU College of Nursing

Author(s):
Minsook Park, PhD - University of Wisconsin at Milwaukee; Geena Kim, Diploma - Chonnam National University;

Systemic Barriers to Transgender and Gender-Nonconforming (TGNC) Assigned Female at Birth (AFAB) Individuals in Accessing Sexual Healthcare

Poster Number: P138
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Diversity, Equity, Inclusion, Accessibility, and Health Equity, User-centered Design Methods, Fairness and Elimination of Bias, Delivering Health Information and Knowledge to the Public
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

We conducted two focus group interviews with 12 transgender and gender-nonconforming (TGNC) assigned female at birth (AFAB) persons regarding their unique experiences in accessing sexual health care. We identified concerns regarding the lack of inclusivity in clinical environments, gender-based discrimination, and sociopolitical tensions influencing physician-patient dynamics and advocates for TGNC inclusion. These findings provided insights to address the systematic barriers to TGNC individuals’ sexual health care and to develop strategies for future TGNC-inclusive initiatives.

Speaker(s):
Vi-Anh Hoang, B.S.
Washington University School of Medicine in St. Louis

Author(s):
Vi-Anh Hoang, B.S. - Washington University School of Medicine in St. Louis; Jennifer Brian, PhD - Arizona State University; Dongwen Wang, PhD - Arizona State University;

Leveraging ChatGPT for Synthesizing Patient Phenotypes in FHIR JSON

Poster Number: P139
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Large Language Models (LLMs), Phenomics and Phenome-wide Association Studies, Interoperability and Health Information Exchange, Natural Language Processing
Primary Track: Foundations
Programmatic Theme: Public Health Informatics

We investigate using ChatGPT 3.5 to create patient phenotypes in FHIR JSON to improve healthcare data management by enhancing interoperability and reducing costs. By leveraging HL7 FHIR patient narratives, we address privacy by generating synthetic patient phenotypes, ensuring compatibility with standard healthcare systems. This approach streamlines data processing, encourages standardized formats, and enhances system efficiency for clinical research and public health informatics. ChatGPT achieves 75% accuracy in generating FHIR JSON phenotypes at no cost.

Speaker(s):
Tia Pope, Ph.D. Student
North Carolina A&T

Author(s):
Ahmad Patooghy, PhD - North Carolina A&T State University;

Launching the AMIA Climate, Health and Informatics Working Group

Poster Number: P140
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Environmental Health and Climate Informatics, Health Equity, Global Health, Population Health, Standards
Working Group: Climate, Health and Informatics Working Group
Primary Track: Applications
Programmatic Theme: Public Health Informatics

AMIA has a unique opportunity to be at the forefront of thought leadership in assessing and addressing the damaging effects of the climate crisis on human health. The Climate, Health and Informatics Working Group was established to support AMIA in its vital role in applying informatics to facilitate research and action addressing climate change and its negative impacts on human health.

Speaker(s):
Titus Schleyer, DMD, PhD
Regenstrief Institute

Author(s):
Michael Zaroukian, MD, PhD, MACP, FHIMSS, ABPM-CI - Self-employed; Manijeh Berenji; Chethan Sarabu, MD - Stanford Medicine; Suzanne Tamang, PhD - Stanford University; Beenish Chaudhry, PhD - University of Louisiana at Lafayette;

Impact of Telehealth on Outpatient Mental Health Utilization Amidst the COVID-19 Pandemic

Poster Number: P141
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Population Health, Telemedicine, Transitions of Care
Primary Track: Foundations
Programmatic Theme: Public Health Informatics

This study examines the impact of COVID-19 on outpatient mental health (OP-MH) utilization among Medicare beneficiaries and emphasizes the role of telehealth in improving OP-MH care access across MH diagnostic conditions. Findings reveal that while patients with schizophrenia, bipolar, and trauma-related disorders returned to pre-pandemic patterns by June 2020, disruptions persisted for those with other MH conditions. Telehealth usage peaked at approximately 60% during the pandemic, gradually declining to ~20% by 2021 across all conditions.

Speaker(s):
Michelle Hayek, Masters of Science
Texas A&M University

Author(s):
Michelle Hayek, Masters of Science - Texas A&M University; Sulki Park - Texas A&M University; Mark Lawley, Ph.D. - Texas A&M University; Michelle Bovin, PhD - National Center for PTSD, VA Boston Healthcare System, Boston, MA, USA; Hye-Chung Kum - Texas A&M University;

Bibliometric Gap Analysis of CDC’s Women’s Health Research, 2018 through 2023: Leveraging National Vital Statistics and Attention Metrics for Insight

Poster Number: P142
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Informatics Implementation, Delivering Health Information and Knowledge to the Public, Health Equity, Fairness and Elimination of Bias
Working Group: Public Health Informatics Working Group
Primary Track: Applications
Programmatic Theme: Public Health Informatics

This analysis identifies conditions with high mortality that disproportionately affect women to inform public health resource allocation. Leveraging informatics methodologies to generate a normalized sex mortality ratio and establish a cause-specific mortality proportion for females, we identified Alzheimer's/dementia and breast cancer as conditions of interest. We then assessed the publication frequency on these conditions in a database of CDC-authored publications and quantified, by topic, the amount of policy citations received by these publications.

Speaker(s):
Victoria Dunkley, MPH
Centers for Disease Control and Prevention (CDC) - Atlanta, GA

Author(s):
Victoria Dunkley, MPH - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Joy Ortega, PhD - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Martha Knuth, MLIS - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Daniel Martin, DrPH - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Christie Kim, BS - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Mary Reynolds, PhD - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Bao-Ping Zhu, MD, PhD, MS - Centers for Disease Control and Prevention (CDC) - Atlanta, GA;

Evaluating the Impact of Electronic Reporting on COVID-19 Data Quality

Poster Number: P143
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Standards, Administrative Systems, Informatics Implementation, Population Health
Primary Track: Applications
Programmatic Theme: Public Health Informatics

The study assessed the impact of a COVID-19 Reporting Portal on data quality attributes, specifically timeliness and completeness. Using laboratory reporting data from January 1 to December 31, 2021, the study compared the new COVID-19 Reporting Tool with the previous fax system. Results showed reductions in the time of test result receipt at the health department for positive and negative results, as well as improvements in data quality for both. Comparing pre- and post-portal implementation data from facilities with both types of data, the tool significantly reduced delays and improved data completeness. These findings highlight the effectiveness of the COVID-19 Reporting Tool in enhancing data quality, which can lead to improved disease surveillance and resource allocation for COVID-19 management. The study concludes that the tool has significantly improved data quality dimensions compared to traditional faxed reporting, offering better information for disease management and preparedness.

Speaker(s):
Donald McCormick, MSHI
Arkansas Hospital Association

Author(s):
Gunes Koru, PhD, FAMIA - University of Arkansas for Medical Sciences, Northwest Regional Campus; Antonije Lazic, MHA - University of Arkansas for Medical Sciences; Haley Barnes, MPH - University of Arkansas for Medical Sciences; Austin Porter, DrPH, MPH - Arkansas Department of Health; Kara Galvan, MPH - Arkansas Department of Health; Clare Brown;

Data Modernization Initiative (DMI) Leadership and Projects across State Public Health Agencies

Poster Number: P144
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Population Health, Informatics Implementation, Interoperability and Health Information Exchange, Infectious Diseases and Epidemiology, Biosurveillance
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Given the multi-billion investments in public health for the Data Modernization Initiative (DMI), it is important to understand the DMI leadership and projects. Analysis pointed to less than half (41%) of DMI leaders having a public health degree, DMI Director designation by one-third (29%) and one-fifth in temporary positions. Analysis of websites pointed to a dedicated DMI site by only one state public health, but all agencies had information on electronic case reporting (eCR) posted.

Speaker(s):
Aasa Dahlberg Schmit, B.Sc.
HLN Consulting

Author(s):
Chanhee Kim, BSN, MPH - University of Minnesota; Pooja Ajmani, BS - University of Minnesota; Sripriya Rajamani, MBBS, MPH, PhD, FAMIA - None;

Analysis of the Public Health Workforce Interests and Needs Survey to Examine Informatics Roles in State and Local Public Health Agencies

Poster Number: P145
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Population Health, Surveys and Needs Analysis, Education and Training, Infectious Diseases and Epidemiology, Informatics Implementation
Working Group: Public Health Informatics Working Group
Primary Track: Applications
Programmatic Theme: Public Health Informatics

The 2021 Public Health Workforce Interests and Needs Survey (PH WINS), a national source of public health workforce was analyzed for current state of public health informatics (PHI) positions. Analysis of 44,732 responses using SAS v9.4 pointed that the PHI specialist workforce accounted for n=144 (1%) of state health agencies, and in informatics program areas (27%) and COVID-19 response (20%). There is a need to strengthen the PHI workforce to data modernization in public health.

Speaker(s):
Divya Rupini Gunashekar, PhD
University of Minnesota- Twin Cities

Author(s):
Harshada Karnik, PhD, MPP - University of Minnesota; Jonathon Leider, PhD - University of Minnesota; Brian Dixon, MPA, PhD - Regenstrief Institute; Sripriya Rajamani, MBBS, MPH, PhD, FAMIA - None;

Comparative analyses of dementia subtypes reveals key node pathways and driver genes enabling drug repositioning

Poster Number: P168
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Drug Discoveries, Repurposing, and Side-effect, Bioinformatics, Systems Biology, Precision Medicine, Omics (genomics, metabolomics, proteomics, transcriptomics, etc.) and Integrative Analyses, Computational Biology, Systems Biology, Data Mining
Primary Track: Applications
Programmatic Theme: Translational Bioinformatics

Introduction: Dementia is a neurodegenerative condition characterized by progressive cognitive decline and functional impairment, that has become a major public health burden with high prevalence in society [1]. The four major subtypes of dementia include: Vascular, Frontotemporal, Pre-senile, and Lewy Body Dementia. Identifying shared and unique genes, pathways and biological network modules across the four subtypes would allow for improved screening, precise diagnosis and developing new treatment options.

Methods: In our research, we developed a computational pipeline (Figure 1(A) analysis to understand the molecular landscape of the different subtypes of dementia to identify unique and shared genetic pathways and driver genes. Disease-gene associations were compiled using the DiSGeNet database. Subtype specific lists were used to perform biological functional enrichment analysis using the KEGG pathway database as the target database and analyzed using Excel, and visualized using Venny.

Results: Based on extensive comparison across all the four subtypes, we have found a core set of 25 genes (Figure 1(B)) driving involved in all subtypes includes granulin precursor (GRN) involved in neuronal development, Apolipoprotein E (APOE) - a key gene involved in neurodegeneration and Superoxide dismutase [Cu-Zn] (SOD1), an antioxidant enzyme involved in reducing cellular toxicity driven by reactive oxygen species - a key component of neurodegenerative diseases. Furthermore, 90 shared pathways (Figure 1(C)) that include insulin resistance, glioma, and hepatitis B. Drug repositioning revealed associations with known drugs targeting neurodegenerative diseases like donepezil and econazole as well as novel associations with antifungals (econazole and butenafine) and immuno-modulatory agent (mepacrine).

Speaker(s):
Zahrahh Shameer, Freshman in High School
Urbana High School

Author(s):

Identifying Patient Segments in Adult Population: Latent Profile Analysis using PROMIS Health Domains and Social Factors from Mixed Data Sources OMIS Health Domains and Social Factors from Mixed Data Sources

Poster Number: P146
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Chronic Care Management, Population Health, Health Equity, Data Mining
Primary Track: Applications
Programmatic Theme: Public Health Informatics

A pragmatic approach to person-centered care is to engage a person’s health domains, such as the ability to participate in physical activities, that are standardized and compared across heterogeneous populations with different chronic conditions. The Patient-Reported Outcome Measurement Information System spans multiple health domains and can be used with electronic health records and neighborhood data to identify characteristics of subgroup analysis. Using K-means clustering and latent profile analysis based on physical function, pain, and mood scores with sociodemographics and health utilization data, we focus on cluster characteristics and quality-of-life predictors. A retrospective cross-sectional study from 2020-2023 was conducted from three physical therapy sites. Outcome variables include sociodemographics, health utilization (eg. access time to care, number of visits), and health status (eg. Charlson comorbidity [CCI] index, body mass index, and PROMIS Physical Function, Pain Interference, and Mood domains. 11,296 patients were analyzed and grouped into four distinct condition profiles: Normal, Mild, Moderate, and Severe symptoms. Patients with Multi-race/Ethnicity, marital status of legally separated, comorbidities like peripheral vascular, cerebrovascular, and peptic ulcer disease, dementia and depression were associated with higher odds of assigned in Severe symptom membership profile compared to the normal group. Higher CCI score and higher ADI scores (indicating more disadvantaged areas) were associated with higher odds of being placed in a Severe symptom profile. This analysis provides further investigations to develop clinical prediction models for change in PROMIS scores and healthcare utilization and identify factors important for the clinical pathways in rehabilitation science.

Speaker(s):
Sang (Sam) Pak, PT, DPT, ACHIP
UCSF

Author(s):
Yuxi Jiang, MS - University of California San Francisco;

Predicting the Information Needs for Domestic Violence Survivors Based on the Fine-tuned Large Language Model

Poster Number: P147
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Natural Language Processing, Information Extraction, Social Media and Connected Health, Data Mining
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Understanding online posts from women facing domestic violence is crucial for timely assistance. We proposed a fine-tuned ChatGPT capable of predicting the informational need of postings, trained with human annotated posts from Reddit. We adopt a progressive strategy to fine-tune GPT-3.5, which achieved 68.75% accuracy, outperforming GPT-3.5 and GPT-4. Our model can rapidly capture the information needs expressed in posts, enabling healthcare providers to provide timely and useful support based on automated predictions.

Speaker(s):
Shaowei GUAN, Bachelor
The Hong Kong Polytechnic University

Author(s):
Vivian Hui, RN, PhD - The Hong Kong Polytechnic University; Rosanna Tsang, Bachelor of Science in Nursing - Hong Kong Polytechnic University; Bohan Zhang, RN, PhD - The Hong Kong Polytechnic University; Arkers Wong, RN, PhD - The Hong Kong Polytechnic University; Rose Constantino, RN, JD, PhD - University of Pittsburgh;

A Large Language Model-Based Voice Interaction Framework for Older Adults Living Alone

Poster Number: P148
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Health Equity, Healthcare Quality, Human-computer Interaction, Self-care/Management/Monitoring, Machine Learning, Natural Language Processing
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Recently, the rapid aging of the population has led to an increase in the number of older adults living alone, leading to loneliness and worsening health issues. We proposed a large language model (LLM) based voice interaction framework to combat loneliness and encourage healthy lifestyles among older adults. This LLM framework with speech technology enables positive interactions, alleviating loneliness, mitigating digital alienation, and healthy lifestyle habits.

Speaker(s):
Kwangsub So, Master
Hallym University

Author(s):
Kwangsub So, ME - Dept. of Artificial Intelligence Convergence, Hallym University; Dong-Soo Shin, RN, PhD - School of Nursing, Hallym University and Center for Primary Care, Harvard Medical School; Jin-Ah Sim, PhD - Dept. of Artificial Intelligence Convergence and College of Medicine, Hallym University; Jae-Jun Lee, MD - College of Medicine, Hallym University; Dong-Ok Won, PhD - Dept. of Artificial Intelligence Convergence and College of Medicine, Hallym University; David Duong, MD, MPH - Center for Primary Care, Harvard Medical School; Kirsten Meisinger, MD, MHCDS - Center for Primary Care, Harvard Medical School;

Analyzing YouTube Videos on Suicide-Related Thoughts and Behaviours: A Study Using Topic Modeling and Discourse Analysis

Poster Number: P149
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Population Health, Natural Language Processing, Deep Learning, Social Media and Connected Health
Primary Track: Foundations
Programmatic Theme: Public Health Informatics

Suicide is a serious public health concern. Youth and young adults are increasingly using social media to share their mental health experiences, including suicide-related thoughts and behaviours (SRTB). Our study examines one popular social media platform, YouTube, to investigate videos on SRTB and their comments, identifying and understanding the topics discussed and how people living with SRTB are portrayed online. We used mixed methods using topic modeling and discourse analysis.

Speaker(s):
Hwayeon Danielle Shin, RN MScN
Institute of Health Policy, Management and Evaluation, University of Toronto/ Centre for Addiction and Mental Health

Author(s):
Jaejun Lee, MSc - No affiliation; Federica Guccini, PhD - Western University; Charles Choi, MD - Department of Psychiatry, University of Toronto; Iman Kassam, MHI - Centre for Addiction and Mental Health, University of Toronto;

Scoping Scientific Publications from the Centers for Disease Control and Prevention, 2014-2023

Poster Number: P150
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Evaluation, Natural Language Processing, Large Language Models (LLMs)
Primary Track: Applications
Programmatic Theme: Public Health Informatics

The Centers for Disease Control and Prevention (CDC) publishes original scientific research to inform public health decision-making. Impacts of the agency’s scientific output can be measured using clinical guidance and policy citations, and news and social media mentions. Large language models provide additional opportunities to assess impact by topic area. We demonstrate use of BERTopic, Altmetric, and BMJ Impact Analytics in conducting an impact analysis of CDC’s publication portfolio.

Speaker(s):
Joy Ortega, PhD
Centers for Disease Control and Prevention (CDC) - Atlanta, GA

Author(s):
Martha Knuth, MLIS - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Christie Kim, BS - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Elissa Meites, MD, MPH - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Mary Reynolds, PhD - Centers for Disease Control and Prevention (CDC) - Atlanta, GA; Bao-Ping Zhu, MD, PhD, MS - Centers for DIsease Control and Prevention (CDC) - Atlanta, GA;

Electronic Case Reporting May Reduce Healthcare Burden

Poster Number: P151
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Documentation Burden, Data Sharing, Informatics Implementation
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Introduction
The Centers for Disease Control and Prevention’s electronic case reporting (eCR) process is a nationwide health information technology solution that automates manual case reporting and may decrease the clinical and administrative burden on healthcare.
Methods
We collected and analyzed labor/cost data for manual case reporting from three healthcare organizations during the study period of March 2020 to March 2021. We also collected the cost to implement and maintain eCR during the same period. We compared the cost for manual case reporting with the cost to implement and maintain eCR during the same period.
Results
During the study period, results from three healthcare organizations demonstrated an average cost of $467,142 labor for 14,043 hours spent manually reporting COVID-19 cases compared with $49,097 labor and 1,343 hours spent implementing and maintaining eCR.
Discussion
Early study results show that electronic case reporting reduced healthcare provider burden and healthcare organization cost compared with manual case reporting. As of November 2024, eCR expanded to 225 conditions beyond COVID-19, allowing providers to spend more time with patients and less time sending burdensome manual case reports.
Reference
Knicely K, Loonsk JW, Hamilton JJ, Fine A, Conn LA. Electronic Case Reporting
Development, Implementation, and Expansion in the United States. Public Health Rep. 2024;0(0). doi:10.1177/00333549241227160

Speaker(s):
Kimberly Knicely, PhD
CDC

Author(s):
Laura Conn, MPH - Centers for Disease Control and Prevention;

Predicting Congestive Heart Failure among Women with Substance Use Disorder Using NIH All of Us Data

Poster Number: P152
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Population Health, Health Equity, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Qualitative Methods, Delivering Health Information and Knowledge to the Public, Machine Learning, Racial Disparities
Primary Track: Applications
Programmatic Theme: Public Health Informatics

This study developed a predictive model for congestive heart failure (CHF) risk in Black and White women with substance use disorders (SUD) using NIH All of Us data. Analyzing records from 2017 to 2023, we found that Black women with SUD had a higher likelihood of CHF compared to White women. Social determinants of health and treatment impact on CHF occurrence, especially among Black women, warrant further investigation for improved health outcomes.

Speaker(s):
Uma Sarder, Masters in Data Science
Meharry Medical College

Uma Sarder, Research Associate
Meharry Medical College, Nashville, TN

Author(s):

Data Integration for Enhanced Movement Behavior Analysis

Poster Number: P153
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Biosurveillance, Data Sharing, Data Transformation/ETL
Primary Track: Applications
Programmatic Theme: Public Health Informatics

The study explores integrating diverse datasets from GPS, WiFi, health, and vaccination records during COVID-19. Despite enrolling 376 individuals, only 61 provided complete integrated data due to challenges like dropouts and non-compliance. This integrated data enables comprehensive analysis of movement patterns, enhances contact tracing, and improves targeted risk communication. Such holistic approaches strengthen public health responses, empowering authorities in combating infectious diseases more efficiently.

Speaker(s):
Sri Surya Krishna Rama Taraka Naren Durbha, Master's in Health Informatics
George Mason University

Author(s):
Janusz Wojtusiak, PhD - George Mason University; Hedyeh Mobahi - College of Health and Human Services, George Mason University;

Understanding information needs in person-centred care for age-related disease, disability and multimorbidity, with emphasis on chronic kidney disease.

Poster Number: P154
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Personal Health Informatics, Aging in Place, Critical Care, Transitions of Care
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Exploring the complexities of delivering person-centred care in multimorbid and disabled populations, especially those with chronic kidney disease, reveals significant gaps in fulfilling the information needs of healthcare professionals and patients. A mixed-methods analysis underscores the critical need for innovative strategies that address these gaps by integrating medical and social needs. Progressing these strategies is crucial for improving care coordination, enhancing health outcomes and better aligning with patients' comprehensive well-being.

Speaker(s):
Jun Wang, Phd student
University of Sheffield

Author(s):

Linking Data Across Clinical and Public Health Systems to Enhance Maternal and Child Health Surveillance in Indiana

Poster Number: P155
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Interoperability and Health Information Exchange, Data Sharing, Population Health, Infectious Diseases and Epidemiology, Pediatrics, Delivering Health Information and Knowledge to the Public
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Indiana has poor maternal and child health (MCH), and surveillance of MCH outcomes is siloed. This poster details an initiative to enhance Indiana's MCH surveillance using data integrated across health care and public health information systems. The project involves enhancement of linkage algorithms for mothers and children; integration of case data; production of surveillance reports; and studies to examine MCH determinants and outcomes. Future efforts will utilize this infrastructure to develop and evaluate MCH interventions.

Speaker(s):
Jill Inderstrodt, PhD/MPH
Indiana University Fairbanks School of Public Health

Author(s):
Jill Inderstrodt, PhD/MPH - Indiana University Fairbanks School of Public Health; Daniel Riggins, MD, MPH - Regenstrief Institute; Acatia Greenwell, BS - Regenstrief Institute; John Price - Regenstrief Institute; Jennifer Crago, MPH - Regenstrief Institute; Stephen O'Brien, BA - Regenstrief Institute; Titus Schleyer, DMD, PhD - Regenstrief Institute; Shaun Grannis, MD - Regenstrief Institute; Brian Dixon, MPA, PhD - Regenstrief Institute;

Exploring Challenges and Opportunities in Government-Academic Data Collaborations for Public Health Crisis Response: A Qualitative Study

Poster Number: P156
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Sharing, Qualitative Methods, Population Health, Legal, Ethical, Social and Regulatory Issues
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Data collaboration between government entities and academia is a promising yet underexplored form of cross-sector partnership in response to public health crises. This poster presents a case study and related interview results regarding a data sharing collaboration between a U.S. state government and a university during COVID-19. We report both organizational and data-specific challenges as well as opportunities, aiming to inform the development of such partnerships for future public health crisis response.

Speaker(s):
Jian-Sin Lee, MSc
University of Michigan School of Information

Author(s):
Tiffany Veinot, PhD - University of Michigan School of Information; Elizabeth Yakel, PhD - University of Michigan School of Information;

The Impact of the mLab App Intervention on HIV Knowledge among Young Cisgender Men and Transgender Women

Poster Number: P157
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Mobile Health, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Delivering Health Information and Knowledge to the Public
Primary Track: Applications
Programmatic Theme: Public Health Informatics

In the United States, young men who have sex with men and young transgender women are disproportionately affected by HIV. Lack of knowledge, early unprotected sexual activity, and low-risk perception among the young population require increased attention. In response to the public health need, this study evaluates the effectiveness of the mLab app intervention in enhancing HIV knowledge over time. The mLab app provides real-time HIV test results and HIV prevention facts and information.

Speaker(s):
Fabiana Dos Santos, PhD, MSN, BSN
Columbia University School of Nursing

Author(s):
Fabiana Dos Santos, PhD, MSN, BSN - Columbia University School of Nursing; Robert Garofalo, MD, MPH - Lurie Children's Hospital; Lisa Kuhns Kuhns, Lisa Kuhns, PhD, MPH - Lurie Children's Hospital; Thomas Scherr, PhD - Vanderbilt University; Rebecca Schnall, RN, PhD - Columbia University;

Combining Word2Vec and Convolutional Neural Network for Emotional Support Classification Among Women with Abusive Experiences

Poster Number: P158
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Deep Learning, Natural Language Processing, Machine Learning, Nursing Informatics, Social Media and Connected Health, Information Extraction, Personal Health Informatics
Primary Track: Applications
Programmatic Theme: Public Health Informatics

This study presents a novel AI application in IPV research, combining Word2Vec and CNN to address challenges in emotional support prediction of IPV-related texts. By using Python programme language to crawl the web and harnessing advanced NLP tools, this study efficiently processes social media data, overcoming the limitations of domain-specific datasets. The findings highlight the model's robustness in emotional support detection and its potential to significantly advance IPV prevention.

Speaker(s):
Yiqing YANG, Bachelor
The Hong Kong Polytechnic University

Author(s):
Rosanna Tsang, Bachelor of Science in Nursing - Hong Kong Polytechnic University; Karen Tsui, Bachelor - Center for Smart Health, School of Nursing, The Hong Kong Polytechnic University; Shaowei GUAN, Bachelor - The Hong Kong Polytechnic University; Arkers Wong, PhD - School of Nursing, The Hong Kong Polytechnic University; Rose Constantino, PhD - Health and community systems, School of Nursing, University of Pittsburgh; Vivian Hui, RN, PhD - The Hong Kong Polytechnic University;

Evaluating the Impact of CHAMPS: A mHealth and Community Health Worker Intervention on Self-Management among People with HIV

Poster Number: P159
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Mobile Health, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Chronic Care Management
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Despite antiretroviral therapy (ART) being widely available in the United States, poor adherence remains a challenge for many patients. To address this challenge, mobile health interventions such as the CHAMPS application have emerged as a promising tool to enhance HIV care, including features to improve ART adherence and self-management among people with HIV (PWH). This study identifies factors influencing the self-management of PWH and evaluates the effectiveness of the CHAMPS intervention on HIV self-management.

Speaker(s):
Fabiana Dos Santos, PhD, MSN, BSN
Columbia University School of Nursing

Author(s):
Fabiana Dos Santos, PhD, MSN, BSN - Columbia University School of Nursing; D. Scott Batey, PhD, MSW - Tulane University/Magic City Research Institute, Birmingham AIDS Outreach; Emma S. Kay, PhD - University of Alabama at Birmingham, Birmingham, AL; Olivia Wood; Joseph A. Abua, MSW - Magic City Research Institute, Birmingham AIDS Outreach, Birmingham; Haomiao Jia, PhD - Columbia University; Rebecca Schnall, RN, PhD - Columbia University;

Collaborative platform to elucidate the significance of plasma cell-free DNA concentration

Poster Number: P160
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Biomarkers, Clinical Decision Support, Data Sharing, Omics (genomics, metabolomics, proteomics, transcriptomics, etc.) and Integrative Analyses, Usability, Bioinformatics
Primary Track: Applications

Plasma cell-free DNA (cfDNA) concentration is expected to be applied in the monitoring of various diseases. We believe that ease of ordering and clarity of result interpretation in clinical practice is necessary to promote its use. We developed the front-end and back-end systems for a measurement service. Not only is this expected to improve sample logistics, but the "clinical workbench" will assist in interpreting patient data using an accumulated anonymized cfDNA data and knowledge base.

Speaker(s):
Teruyoshi Hishiki, MD
Toho University / Crecon Genomics Inc. /

Author(s):
Joanna Noack, MS - Crecon Genomics Inc.; Shuntaro Tamura, MS - Crecon Genomics Inc.; Takuro Tamura, MS - Research and Development Center for Precision Medicine, University of Tsukuba / Crecon Genomics Inc. /; Shinji Irie, Ph.D - iLAC Co., Ltd.; Toshiaki Ito, MS - Crecon Genomics Inc.;

HydraNet: a Causal AI Model for Precision Immune Therapy Balancing Survival Benefits and Severe Immune Adverse Event Risk

Poster Number: P161
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Causal Inference, Real-World Evidence Generation, Deep Learning, Precision Medicine, Drug Discoveries, Repurposing, and Side-effect
Primary Track: Applications

Immune therapies have transformed cancer treatment, yet severe immune-related adverse events remain concerning. We developed HydraNet, a novel causal AI model, to elucidate the tradeoff between survival benefits and severe AKI risks across regimens involving immune checkpoint inhibitors. HydraNet leverages generalized propensity scores in the targeted regularization to enable causal inference for multiple treatments and enhance model stability. HydraNet performance is compared with multivariate regression models and inverse probability weights using N3C EMR data.

Speaker(s):
Yao Chen, Master of Science
Indiana University School of Medicine

Author(s):
Vithal Madhira - Palila Software; Xiaochun Li, PhD - Indiana University School of Medicine; Tyler Shugg, PharmD, PhD - Indiana University School of Medicine; Shadia Jalal, MD - Indiana University Melvin and Bren Simon Comprehensive Cancer Center; Fang-Chi Hsu, PhD - Wake Forest School of Medicine; Michael Eadon, MD - Indiana University School of Medicine; Benjamin Bates, MD - Department of Medicine, Rutgers-RWJMS Medical School; Noha Sharafeldin, MD, PhD, MSc - School of Medicine, University of Alabama at Birmingham, Birmingham; Umit Topaloglu, PhD - National Cancer Institute; Qianqian Song, Ph.D. - University of Florida; Jing Su, PhD - Indiana University School of Medicine;

Deep Learning Approach to Predict Lung Adenocarcinoma Recurrence

Poster Number: P162
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Deep Learning, Machine Learning, Cancer Genetics, Cancer Prevention, Bioinformatics, Omics (genomics, metabolomics, proteomics, transcriptomics, etc.) and Integrative Analyses
Primary Track: Applications
Programmatic Theme: Translational Bioinformatics

Our project examines the ability of Deep Learning (DL) models to predict Lung Adenocarcinoma (LUAD) recurrence (data from the LUAD TCGA PanCancer Atlas dataset). Our dataset included clinical features (age, sex, tumor type, clinical (TNM) and pathologic AJCC tumor staging), mrna expression Z-scores, non-synonymous mutation counts per gene, and gene methylation data. When compared to machine learning, DL model achieved high AUROC and AUPRC, and gave insights into the features most involved in LUAD recurrence.

Speaker(s):
Karma Hayek, BS
Brown University

Author(s):
Karma Hayek, BS - Brown University; Jessica Patricoski, PhD student, Computational Biology - Brown University Center for Computational Molecular Biology; Jeremy Warner, MD, MS - Brown University; Ece Uzun, PhD - Lifespan/Brown University;

A High-Throughput Approach to Generating Comorbidity Scores for Phenotypes

Poster Number: P163
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Phenomics and Phenome-wide Association Studies, Bioinformatics, Population Health
Primary Track: Foundations
Programmatic Theme: Translational Bioinformatics

Comorbidities are an important source of confounding in observational studies; however, properly controlling for comorbidities related to a disease phenotype can be challenging. We propose a data-driven, high-throughput approach to generating comorbidity scores for phenotypes using electronic health record data. We perform phenome-wide association studies to identify conditions that are significantly positively associated with specific phenotypes. Our preliminary results reveal associations of schizophrenia and Alzheimer’s disease and of chronic ulcers and type 1 diabetes mellitus.

Speaker(s):
Monika Grabowska
Vanderbilt University

Author(s):
QiPing Feng, PhD - Vanderbilt University Medical Center; Wei-Qi Wei, MD, PhD - Vanderbilt University;

A Network Biology-based Framework for Identifying Potential Drug Targets for Glioblastoma

Poster Number: P164
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Cancer Genetics, Drug Discoveries, Repurposing, and Side-effect, Computational Biology, Omics (genomics, metabolomics, proteomics, transcriptomics, etc.) and Integrative Analyses, Systems Biology
Primary Track: Applications
Programmatic Theme: Translational Bioinformatics

This study focuses on Glioblastoma (GBM), the most common malignant brain tumor in the central nervous system, representing a significant portion of malignant CNS tumors. GBM is characterized by its rapid progression, leading to a median overall survival time of 12-15 months despite the primary treatment protocol of surgical removal, radiotherapy, and chemotherapy. The research emphasizes the complexity of GBM, particularly its genetic diversity and protein-protein interaction (PPI) network disruptions. Utilizing 'omics' research, the study highlights the genetic and transcriptomic variability in GBM tumors as a critical challenge in overcoming treatment resistance and developing universally effective therapies. Through the innovative use of the Proteinarium tool for PPI network analysis, this research demonstrates GBM's molecular heterogeneity, identifying clusters of patients with shared PPI networks. The analysis, based on data from The Cancer Genome Atlas (166 patients) and validated with the Chinese Glioma Genome Atlas (85 patients), identified three subgroups of GBM patients with distinct molecular and clinical profiles. Notably, one subgroup showed significantly lower survival rates, with key genes associated with inflammatory response and angiogenesis identified as potential drug targets. This study offers new avenues for GBM treatment by targeting specific proteins within the consensus PPI network.

Speaker(s):
Alper Uzun, PhD
Brown, University, Department of Pathology and Laboratory Medicine, Department of Pediatrics

Author(s):
Charissa Chou - Brown University; Jennifer Li, NA - Brown Universiy; Sean Lawler, PhD - Brown University; Ece Uzun, PhD - Lifespan/Brown University; Alper Uzun, PhD - Brown, University, Department of Pathology and Laboratory Medicine, Department of Pediatrics;

Development of a Body Movement Classification Algorithm Using Wearable Accelerometers: A Pilot Study

Poster Number: P165
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Machine Learning, Deep Learning, Ubiquitous Computing and Sensors
Primary Track: Foundations
Programmatic Theme: Translational Bioinformatics

Human activity recognition via wearable sensors holds promise for patient monitoring in multiple hospital and outpatient settings. We conducted tests with 20 participants wearing accelerometers, analyzing 17 different activities. Data preprocessing, transformation, feature engineering, and model implementation were performed. Results showed an average 94% accuracy in classifying tasks. This study establishes a model for automated activity recognition with potential clinical implications, demonstrating the effectiveness of integrating accelerometers and machine learning algorithms for real-time monitoring.

Speaker(s):
Juan Garcia-Mendez, M.D.
Mayo Clinic

Author(s):
Vitaly Herasevich, MD, PhD, FCCM, FAMIA - Mayo Clinic; MOHAMMAD JOGHATAEE, PhD - Auburn University; Ashish Gupta, PhD - Auburn University; Juan Garcia-Mendez, M.D. - Mayo Clinic; Kushagra Kushagra, PhD - Auburn University;

Molecular Subtypes of Ovarian Cystadenocarcinomas Using Protein-Protein Interaction Networks

Poster Number: P167
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Biomarkers, Cancer Genetics, Omics (genomics, metabolomics, proteomics, transcriptomics, etc.) and Integrative Analyses
Primary Track: Applications
Programmatic Theme: Translational Bioinformatics

Ovarian malignancies have a very poor survival rate which can be low as 31.5% when diagnosed at a metastatic stage. With our study, we have identified unique genes which harbor the potential for future targeted therapies by analyzing RNA-sequencing data from The Cancer Genomic Atlas (TCGA) Firehose Legacy data of 600 patients. To achieve this, we utilized a multi-sample tool (Proteinarium) to analyze and create patient clusters with unique protein-protein interactions.

Speaker(s):
Ece Uzun, PhD
Lifespan/Brown University

Author(s):
Jessica Claus, M.D.; Amelia Zug, N/A - Brown University; Joyce Ou, MD, PhD - Brown University; Alper Uzun, PhD - Brown, University, Department of Pathology and Laboratory Medicine, Department of Pediatrics; Ece Uzun, PhD - Lifespan/Brown University;

Are Health Care Students Ready For AI?

Poster Number: P07
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Education and Training, Surveys and Needs Analysis, Curriculum Development
Primary Track: Applications

Purpose: Proficiency in comprehending the impact of AI technology on clinical work-flow and the ability to interpret AI interventions are essential practical competencies. There is limited research assessing students’ learning needs, fears, and anxieties related to learning about AI technology in health care. Addressing this research gap is important to gain understanding of how best to integrate AI education into curricula.
Method: This multi-Institutions project uses a cross-sectional design using an electronic survey to conduct a needs assessment capturing insights from medical, nursing, and pharmacy students’ regarding the significance of acquiring competencies in AI applications within healthcare. The survey consists of 3 sections including: 1)Assessing students’ perception of the importance of AI competency integration in the curriculum; 2) surveying students’ AI learning anxiety/fears using questionnaire adapted from a validated AI anxiety tool; and 3) demographic information of students and digital use self-efficacy.
Evaluation: Descriptive statistics will be used to summarize demographic data. Each outcome (importance of competency integration for each competency, digital use self-efficacy and AI anxiety) is scored by averaging associated items to form a scale. Fitting linear regression model will be used to predict the outcomes (importance of competency integration for each competency, digital fluency, and AI anxiety) by students.
Conclusion: The study data will provide valuable insights into the perspectives of medical, nursing, and pharmacy students regarding the integration of AI competencies into their professional education. Findings from the study will help faculty tailor their curricula and identify support intervention for those learners’ with AI anxiety.

Speaker(s):
Linda Chang, PharmD, MPH
University of Illinois Chicago

Author(s):

Identification of rare cell population and improving single-cell clustering accuracy using the RECOMBINE marker selection algorithm

Poster Number: P166
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Omics (genomics, metabolomics, proteomics, transcriptomics, etc.) and Integrative Analyses, Computational Biology
Primary Track: Applications
Programmatic Theme: Translational Bioinformatics

In this study, we apply RECOMBINE, an innovative algorithm enhancing marker selection for scRNA-seq data, optimizing the detection of rare cell populations. Leveraging sparse hierarchical clustering with added penalties, RECOMBINE outperforms existing methods, as evidenced by superior Adjusted Rand Index scores. Our analysis demonstrates its efficacy in discerning distinct cell types within complex datasets, promising advancements in cellular classification and insights into intricate biological systems.

Speaker(s):
Wanru Guo, BPharm(Hons), MS
UTHealth

Author(s):

Domain-specific LLM Development and Evaluation – A Case-study for Prostate Cancer

Poster Number: P169
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Large Language Models (LLMs), Deep Learning
Primary Track: Applications

In this work, we present our efforts for development of domain-specific large language models. Prostate cancer was chosen as a use-case for this study. We collected more than 5 million clinical notes and radiology and pathology reports for XX patients treated for prostate cancer in Mayo Clinic. In addition to domain-specific training data, we built domain-specific tokenizers and devised domain-specific training strategies for LLM development. LLM was forced to focus on domain-specific information by marking clinical terms using UMLS parser. UMLS parser was used to identify treatment, symptoms, side effects, medication, and anatomical details in text snippets. Model was trained to predict these terms using the context of surrounding text. We evaluated the model for downstream tasks of clinical information prediction and question answering in comparison to a similarly sized general purpose model GPT-2 and a three-times larger domain specialized model. i.e., BioGPT. Our model outperformed GPT-2 on both tasks by a wide margin. Our model was also able to outperform BioGPT on clinical information prediction task and showed some advantages over BioGPT in question-answering task.

Speaker(s):
Avisha Das, Ph.D.
Mayo Clinic

Author(s):
Amara Tariq, Ph.D. - Mayo Clinic Arizona; Man Luo, PhD - Mayo Clinic Arizona; Aisha Urooj, Ph.D. - Mayo Clinic Arizona; Avisha Das, Ph.D. - UTHealth Science Center Houston; Jiwoong Jeong, M.S. - Arizona State University; Shubham Trivedi, B.S. - Mayo Clinic Arizona; Bhavik Patel, MD, MBA - Mayo Clinic;

A ChatGPT Empowered Method with MicroRNA-Gene-Disease Knowledge Graph in Biomedical Research

Poster Number: P170
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Cancer Prevention, Biomarkers, Disease Models
Primary Track: Applications

Accurately identifying the relationship between regulators and diseases has long been a challenge. However, understanding precise causal pathways remains challenging, primarily due to dispersed data annotation across databases. This calls for a unified data visualization format to streamline these resources, aiding the identification of biomarkers and therapeutic targets. This study addresses these challenges by developing the first unified miRNA-gene-disease knowledge graph. A case study for this application focused on COVID-19 identified impacts on fetal growth.

Speaker(s):
Judy Bai, High school student
Greenhills School

Author(s):

Comparing EHR-recorded Race/Ethnicity to Self-reported Race/Ethnicity: Insights from the All of Us Research Program

Poster Number: P171
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Data Sharing, Information Visualization
Primary Track: Applications

This study evaluates the accuracy of EHR-recorded race/ethnicity against the gold standard of self-reported data from the All of Us Research Program, aiming to identify factors on accuracy across various groups. Performance metrics such as sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated to evaluate EHR-recorded race/ethnicity data against participant self-reported race/ethnicity separately. The logistic regression was used to examine the association between race/ethnicity agreement (agree/disagree) and EHR-recorded race/ethnicity, also the association between EHR race/ethnicity missing (yes/no) and self-reported race/ethnicity.

Results from 217,546 participants show a 12% discrepancy rate between EHR and self-reported race, with high accuracy for Whites. Non-White race groups were more likely to have race discrepancies and missing EHR race data. Out of 224691participants in the All of Us program with recorded ethnicity data in both EHR and the Basic survey, 4% reported a different ethnicity in the EHR compared to the survey. Hispanic participants were more likely to report ethnicity discrepancies.

Our study reveals high concordance of EHR-recorded race with self-reported race for White participants, followed by Black and Asian individuals. However, significant discrepancies exist for other race groups, including MENA, NHPI, Others, and Mixed. This underscores the necessity for efforts to enhance the accuracy of EHR-recorded race/ethnicity for certain underrepresented groups.

Speaker(s):
Lina Sulieman, PhD
Vanderbilt University Medical Center

Author(s):

Systemic collection, annotation, and analysis, of viral vaccines in the VIOLIN database

Poster Number: P172
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Bioinformatics, Data Mining, Disease Models
Primary Track: Applications
Programmatic Theme: Academic Informatics / LIEAF

Viral vaccines have been proven significant in protecting us against viral diseases such as COVID-19. We systematically collected, manually annotated, and analyzed 2,847 viral vaccines against 95 viral species, and stored the information of the vaccines, vaccine components such as 542 vaccine antigens, vaccine formulations, and their induced host responses in the VIOLIN vaccine database. Enriched patterns (e.g., viral entry into the host cell, DNA binding or RNA binding) were obtained from our Gene Ontology enrichment analysis using a customized Fisher’s exact test. Unexpectedly , only a small percentage of the vaccine antigens from 74 RNA and 20 DNA viral species were predicted to be adhesin, sugges ting no strong correlation between protective viral antigens and adhesin probability. The viral vaccines and their associated entities and relations are also ontologically modeled and represented in the Vaccine Ontology. A VIOLIN web interface was developed to support user friendly queries of viral vaccines.

Speaker(s):
Anthony Huffman, Bioinformatics
University of Michigan

Author(s):

Breast cancer family history in participants with BRCA variants: A study in All of Us Research Program

Poster Number: P173
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Cancer Genetics, Omics (genomics, metabolomics, proteomics, transcriptomics, etc.) and Integrative Analyses, Informatics Implementation, Biomarkers
Primary Track: Applications
Programmatic Theme: Translational Bioinformatics

The All of Us Research Program is creating the most diverse biomedical research repository and aims to establish return of results program to identify participants at risk of diseases (e.g., breast cancer). We combined multi-omic (survey, EHR, genomics) to assess the BRCA prevalence in participants with breast cancer family history. Participants with breast cancer family history and BRCA genes had higher breast cancer incidence. Expanding and enhancing analysis can be first step in returning results to participants.

Speaker(s):
Lina Sulieman, PhD
Vanderbilt University Medical Center

Author(s):

The Burden of Cancer and Pre-cancerous Conditions Among Transgender Individuals in a Large Healthcare Network

Poster Number: P174
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Cancer Prevention, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Real-World Evidence Generation
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The current study aimed to examine the prevalence of and risk factors for cancer and pre-cancerous conditions, comparing transgender and cisgender individuals, using 2012-2023 electronic health record data from a large healthcare system. We identified 2,745 transgender individuals using a previously validated computable phenotype and 54,900 matched cisgender individuals. We calculated the prevalence of cancer and pre-cancer related to human papillomavirus (HPV), human immunodeficiency virus (HIV), tobacco, alcohol, lung, breast, colorectum, and built multivariable logistic models to examine the association between gender identity and the presence of cancer or pre-cancer. Results indicated similar odds of developing cancer across gender identities, but transgender individuals exhibited significantly higher risks for pre-cancerous conditions, including alcohol-related, breast, and colorectal pre-cancers compared to cisgender women, and HPV-related, tobacco-related, alcohol-related, and colorectal pre-cancers compared to cisgender men. These findings underscore the need for tailored interventions and policies addressing cancer health disparities affecting the transgender population.

Speaker(s):
Shuang Yang, MS
University of Florida

Author(s):
Shuang Yang, MS - University of Florida; Yongqiu Li; Christopher Wheldon; Mattia Prosperi, PhD, FAMIA - University of Florida; Thomas George, MD, FACP - Division of Hematology and Oncology, University of Florida, Gainesville, Florida, USA;; Elizabeth Shenkman, PhD - University of Florida Health; Fei Wang, PhD - Weill Cornell Medicine; Jiang Bian, PhD - University of Florida; Yi Guo, PhD - University of Florida;

Impact of Stressful Life Events on Preventive Colon Cancer Screening

Poster Number: P175
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Cancer Prevention, Large Language Models (LLMs), Population Health, Real-World Evidence Generation
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Stressful life events are important psychosocial predictors of health. In this project we use large language models to extract information about such events from health records. We were able to show that some events, such as divorce, may act as facilitators of screening, while others, such as arrest, act as barriers.

Speaker(s):
Paul Heider, PhD
Medical University of South Carolina

Author(s):
Leslie Lenert, MD - Medical University of South Carolina; Jihad Obeid, MD - Medical University of South Carolina; Ramsey Wehbe, MD, MSAI - Medical University of South Carolina; Dmitry Scherbakov, PhD - Medical University of South Carolina;

Revolutionizing Postoperative Ileus Monitoring: Exploring GRU-D's Real-Time Capabilities and Cross-Hospital Transferability

Poster Number: P176
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Deep Learning, Critical Care
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Postoperative ileus (POI) poses a common challenge following colorectal surgery, contributing to heightened morbidity, increased costs, and prolonged hospital stays. While studies have discussed the prediction of POI under traditional statistical frameworks, there exists a notable gap concerning the performance of deep learning-based approaches in POI prediction. Here we explored the performance and transferability of GRU-D based deep learning architecture for real-time risk modeling of postoperative ileus. Our study indicates strong transferability of the deep learning model across hospital sites and electronic health record systems with non-overlapping surgery date frames. Along with the ability to automate missing parameterization and manage irregular sampling data, the proposed architecture inches closer to real-world clinical practice.

Speaker(s):
Xiaoyang Ruan, PhD
The University of Texas Health Science Center at Houston

Author(s):
Sunyang Fu, PhD, MHI - UTHealth; Kellie Mathis, M.D. - Mayo Clinic; Cornelius Thiels, M.B.A - Mayo Clinic; Patrick Wilson, PhD - Mayo Clinic; Curtis Storlie, PhD - Mayo Clinic; Hongfang Liu, PhD - University of Texas Health Science Center at Houston; Xiaoyang Ruan, PhD - The University of Texas Health Science Center at Houston;

Multi-Sensor Monitoring of Surgical Nociception

Poster Number: P177
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Machine Learning, Self-care/Management/Monitoring, Critical Care, Ubiquitous Computing and Sensors, Surgery, Internal Medicine or Medical Subspecialty, Precision Medicine
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Objectively monitoring a patient’s unconscious pain state during surgery remains a challenge, contributing to postoperative consequences including delayed healing and cognitive dysfunction. In a prospective study of 101 surgeries containing ~50000 surgical pain stimuli, we demonstrate dynamic tracking of unconscious pain processing throughout surgery using supervised and unsupervised machine learning frameworks. Our findings show that a multi-sensor approach identifies a physiologically consistent pain signature. This signature represents a principled AI-guided approach to inform anesthetic dosing.

Speaker(s):
Sandya Subramanian, Ph.D.
UC Berkeley

Author(s):
Sandya Subramanian, Ph.D. - UC Berkeley; Bryan Tseng, B.S. - Massachusetts Institute of Technology; Marcela Del Carmen, M.D., M.P.H. - Massachusetts General Hospital; Annekathryn Goodman, M.D., M.S., M.P.H. - Massachusetts General Hospital; Douglas Dahl, M.D. - Massachusetts General Hospital; Riccardo Barbieri, Ph.D. - Politecnico di Milano; Emery Brown, M.D., Ph.D. - Massachusetts Institute of Technology / Massachusetts General Hospital;

Early Peripheral Vascular Intervention for Treating Patients with Claudication: A Decision Analysis using National Administrative Claims Data

Poster Number: P178
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Surgery, Healthcare Quality
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The objective of this study was to evaluate the impact of early peripheral vascular interventions (PVI) versus conservative management on the management of claudication in patients with peripheral arterial disease (PAD). Utilizing a decision model, we assessed the trade-offs between early PVI and conservative management considering various clinical outcomes. Data were obtained from Medicare fee-for-service claims, and probabilities were calculated based on patient encounters for claudication, PVI, open bypass, progression to Chronic Limb Threatening Ischemia (CLTI), and lower extremity amputation. A Markov model was constructed to simulate patient outcomes over a 2-year period.

Results indicated that patients undergoing early PVI experienced more PVIs and open bypass procedures compared to those receiving conservative management. Furthermore, the duration of time spent in a state of claudication was shorter for patients who underwent early PVI. These findings suggest that early PVI may not necessarily lead to improved outcomes and may even result in adverse effects, requiring additional interventions.

In conclusion, our study highlights the importance of leveraging healthcare informatics to inform decision-making processes in managing claudication among Medicare beneficiaries. By utilizing claims data and analytical methodologies, we can optimize resource allocation and promote evidence-based practices in healthcare delivery. These insights underscore the significance of integrating data-driven approaches to enhance patient care and outcomes in the management of PAD-related symptoms like claudication.

Speaker(s):
Chen Dun, MHS
Johns Hopkins University

Author(s):
Caitlin Hicks, MD, MS - The Johns Hopkins School of Medicine; Harold Lehmann, MD, PhD - Johns Hopkins University;

Using Structured EHR Data to Develop a Breast Cancer Risk Estimator

Poster Number: P179
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Machine Learning, Cancer Prevention, Population Health
Primary Track: Applications

Breast cancer is the most diagnosed cancer among women and early detection improves 5-year survival rate from 31.0% to 99.3%. This study presents a Breast Cancer Risk Estimator (BCRE) classifier that predicts the 6-month likelihood of breast cancer diagnosis based on routine EHR data, without imaging or additional questionnaire collected from the patient. The proposed BCRE classifier can enable population-level risk stratification to be performed easily, facilitating timely outreach for potentially life-saving screening and treatment.

Speaker(s):
Tamanna Tabassum Munia, PhD
Geisinger

Author(s):
Tamanna Tabassum Munia, PhD - Geisinger; Timothy Murphy, MD - Geisinger; Rosemary Leeming, MD, MHCM - Geisinger; Robin Skrine, MD - Geisinger; Ginger Hill, R.T., (R)(M) - Geisinger; Elliot Mitchell, PhD - Geisinger; Rebecca Maff, MS; Casey Cauthorn, MIE - Geisinger; David Vawdrey, PhD - Geisinger; Aalpen Patel - Geisinger Health System; Biplab S Bhattacharya, PhD - Geisinger Health System;

External Validation of a Clinically Validated Machine Learning Model to Identify and Prevent Acute Care During Cancer Radiotherapy and Chemoradiation

Poster Number: P180
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Machine Learning, Precision Medicine, Chronic Care Management
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Patients with cancer undergoing radiation therapy (RT) often receive acute care during treatment. The SHIELD-RT algorithm uses structured electronic health record data to identify patients at high risk for acute events during RT and demonstrated high predictive performance in our previous randomized controlled trial. The objective of this study was to perform an external evaluation of algorithmic performance at a large academic medical center.

Speaker(s):
Marianna Elia, PhD
UCSF

Author(s):

Fairness in Early Warning Scores

Poster Number: P181
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support, Fairness and elimination of bias, Racial disparities, Critical Care
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Hospitals use early warning scores (EWS) to detect patients at risk of critical deterioration and intervene to rescue them. Here, we evaluate three EWS systems for algorithmic fairness. Notably, our study utilizes a novel evaluation framework to discover that EWS usage could lead to differential true positive rates across racial and ethnic subgroups of patients, independent of score thresholds and the capacity to allocate.

Speaker(s):
Anoop Mayampurath, PhD
University of Wisconsin - Madison

Author(s):

Enhancing Medical Guideline Retrieval with Language Model-Augmented Retrieval (LMAR): A Comparative Study with Traditional RAG

Poster Number: P182
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Guidelines, Clinical Decision Support, Large Language Models (LLMs)
Primary Track: Applications
Programmatic Theme: Clinical Informatics

The development of retrieval-augmented generation (RAG) methods marked a significant advancement in natural language processing, particularly in clinical decision support systems. However, traditional RAG methods often struggle with accurately retrieving contextually relevant information from medical guidelines. Our study introduces Language Model-Augmented Retrieval (LMAR), an innovative approach that leverages large language models to enhance the retrieval and generation of information, ensuring the provision of accurate and actionable guidance.

We conducted a comparative study using a dataset of 42 clinical guidelines. The study compared traditional RAG, which uses cosine similarity for retrieval, with LMAR, which employs a prompted large language model to rank the relevance of each guideline based on metadata and the query. The effectiveness of each method was evaluated based on its ability to accurately retrieve dosing information for pharmacological treatments in response to preselected clinical queries.

Traditional RAG demonstrated limitations, often retrieving non-actionable information. In contrast, LMAR was able to capture medication dosing information in 100% of cases, accurately retrieving information from sections and tables within the guidelines. Our study demonstrates that LMAR significantly outperforms traditional RAG in extracting actionable guidance from medical guidelines.

Speaker(s):
Joongheum Park, MD
Chobanian & Avedisian School of Medicine, Boston University

Author(s):

Enhancing Health Evidence Quality in Classification Tasks: A Triangulation Approach Utilizing Case-Based Reasoning and Process Feature

Poster Number: P183
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Mining, Real-World Evidence Generation, Knowledge Representation and Information Modeling
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

For a given classification task, a model's performance ceiling usually depends on whether unknown information includes deterministic factors linked to the desired outcome. This study proposed a triangulation approach to enhance the quality of health insights derived from such tasks. The framework synthesizes multidimensional information from local process mining and case-based reasoning, leading to a dramatic reduction in the proportion of misclassified cases from 38.5% to 9.8% in an ischemic heart disease classification task.

Speaker(s):
Simon Poon, PhD
University of Sydney

Author(s):
Ruihua GUO, PhD Candidate - School of Computer Science, The University of Sydney; Ross Smith, MBBS, PhD Candidate - The University of Sydney; Qifan Chen, Doctor of Philosophy - School of Computer Science; Angus Ritchie, MBBS FRACP FACHI - Sydney Local Health District;

Elucidating The Impact of Community-Level Social Determinants of Health on Pre-operative Frailty: A Data-Driven Study in Florida

Poster Number: P184
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Mining, Machine Learning, Aging in Place
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Frailty, an age-related syndrome, is associated with poor post-operative outcomes. The impact of community-level social determinants of health (SDoH) on pre-operative frailty has not been investigated yet. We developed a machine learning model to predict pre-operative frailty using an institutional dataset and applied it to a more geographically diverse population from the OneFlorida+ Clinical Research Consortium. Computable phenotyping for SDoH stratification using unsupervised learning was employed to identify distinct patient profiles based on zip code-level SDoH characteristics. We applied multivariate logistic regression to examine the association between SDoH profiles and pre-operative frailty risk. Adverse community-level SDoH profiles are independently associated with higher pre-operative frailty risk; patients from the disadvantaged SDoH profile had 1.21 times higher odds (95% CI 1.16-1.26) of being frail compared to the advantaged SDoH cluster after adjusting for potential confounders. Considering patients’ social context could improve pre-operative care and surgical outcomes, informing clinical practice and policies.

Speaker(s):
Mamoun Mardini, PhD
University of Florida

Author(s):
Chen Bai, MS - University of Florida;

Trends in Randomized FDA Registration Trial Design in Hematology and Oncology Trials Over Time

Poster Number: P185
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Mining, Healthcare Quality, Internal Medicine or Medical Subspecialty, Drug Discoveries, Repurposing, and Side-effect
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

The predominant trial design for FDA approval in emerging oncological therapies is generally an escalatory approach. In contrast to introducing novel therapeutic schemes, these studies incorporate the investigated drug into existing regimens. Our analysis, spanning October 2004 to the present, not only shows the increase of escalatory design in general but highlights a need to consider trial design as a crucial factor in evaluating FDA approvals.

Speaker(s):
Aidan Petrovich, MD
Northside Hospital

Author(s):
Aidan Petrovich, MD - Northside Hospital; Jeremy Warner, MD, MS - Brown University; Wayne Liang, MD MS FAMIA - Children's Healthcare of Atlanta & Emory University; Samuel Rubinstein, MD, MSCI - University of North Carolina-Chapel Hill School of Medicine;

Identifying Pediatric Intensive Care Unit Bedside Teams and their Stability Using Electronic Health Record Audit Log Data

Poster Number: P186
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Mining, Healthcare Quality, Machine Learning, Information Retrieval, Workflow, Administrative Systems, Critical Care, Clinical Decision Support
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Inpatient care teams often consist of varying combinations of clinicians that are not explicitly documented and unique to patients, making them difficult to identify. This may be beneficial, but introduces risk of impaired communication and diminished trust among unfamiliar team members. Current methods of studying inpatient teams are too effort-intensive to develop actionable insights at scale.

We propose two methods for identifying and measuring the stability of patient-centric teams using EHR audit logs: (1)a rule-based, heuristic approach; (2)an unsupervised clustering approach.

We collected audit log data from 194 pediatric intensive care unit (PICU) patient encounters and 3121 clinicians from calendar year 2022 at a large academic children’s hospital. We used these data to identify clinician-teams directly involved in each patient's care each day of their encounter. For the rules-based approach, we identified clinicians with specific audit log actions indicative of bedside presence. For the clustering-based approach, we developed features that captured proportionality of audit log actions performed, embedded them to a 2D vector, performed clustering, and selected team members based on clusters. We computed stability scores for team compositions identified by these approaches.

We successfully identified team compositions that yielded stability scores aligned with clinical expectations. We demonstrated the feasibility of using audit logs to identify and characterize inpatient teams at scale. This methodology enables more complex team-related measures that lead to high/low-quality inpatient care to be developed and studied. Our next steps are to conduct a direct observation study to validate these measures against a gold standard.

Speaker(s):
Liem Nguyen, Undergraduate
Stanford University

Author(s):
Liem Nguyen, Undergraduate - Stanford University; Seunghwan Kim, MS - Washington University in St. Louis; Thomas Kannampallil, PhD - Washington University School of Medicine; Stefanie Sebok-Syer; Daniel Tawfik, MD, MS - Stanford University School of Medicine;

A Longitudinal Analysis of Institutional Adoption, Use, and Dissemination of an EHR Vendor-Based Data Sharing Program

Poster Number: P187
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Data Sharing, Real-World Evidence Generation, Administrative Systems
Primary Track: Applications
Programmatic Theme: Clinical Informatics

While health systems increasingly participate in real-world data sharing collaboratives, little is known about their empiric use. To examine one proprietary research collaborative in detail, administrative data from Cosmos research collaborative (Epic Corporation) was analyzed over 27 months through the end of 2023. An increasing number of organizations participated in Cosmos across geographic regions and organization types between Quarter 4 2021 and end of 2023 (152 to 229 total organizations “live” on Cosmos). While distinct users increased 3-fold over this time period, user engagement remained on average low (between 0.25 to 0.21 projects on average per user per quarter). There was a trend toward increased number of logins and time using the platform over the study period with 27 total publications referencing Cosmos (25 with Cosmos data analysis). Although adoption is increasing for Cosmos, opportunities remain to improve cross-organizational data collaborative engagement for this and other similar platforms.

Speaker(s):
Melissa Gunderson, MD
University of Minnesota

Author(s):
David Dorr, MD - Oregon Health & Science University; Harry Freedman - Epic; Genevieve Melton-Meaux, MD, PhD - University of Minnesota;

Investigating Privacy and Generalization in Deep Survival Analysis Through Modeling SEER Metastatic Prostate Cancer Patient Outcomes

Poster Number: P188
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Deep Learning, Privacy and Security, Internal Medicine or Medical Subspecialty
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

The application of deep learning to survival analysis frequently improves model performance, with the risks of overfitting and information leakage. Here we report the performance metrics of traditional and deep survival models trained on metastatic prostate cancer patient data from the SEER program database, as well as the same metrics with the addition of differential privacy. We specifically focus on how these models generalize to settings of racially homogenous versus diverse test sets.

Speaker(s):
Joseph Vento, MD
Vanderbilt University Medical Center

Author(s):
Luca Bonomi, PhD - Vanderbilt University Department of Biomedical Informatics;

Oncology Decision Support in Ovarian Cancer: Artificial Intelligence-Based Pathomics to Identify Platinum-Resistant Epithelial Ovarian Cancer

Poster Number: P189
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Deep Learning, Machine Learning, Clinical Decision Support, Precision Medicine
Primary Track: Applications
Programmatic Theme: Clinical Informatics

People with ovarian cancer typically present at an advanced stage, and the standard of care is extensive surgery followed by platinum-based chemotherapy. However, approximately 15% of people will be platinum-resistant, meaning that they will relapse within six months of their surgery. Such patients require different, more aggressive chemotherapy than the standard, which could increase their time to progression and survival rate. To this end, we propose a multimodal deep learning framework to identify people with platinum-resistant epithelial ovarian cancer and resistance to bevacizumab, a recently FDA approved targeted molecular therapy for ovarian cancer. In particular, we take into account not only gigapixel histopathology images, but also patient clinical variables under a multi-modal learning framework. Evaluation is performed on two ovarian cancer benchmarks: a public The Cancer Imaging Archive Ovarian Bevacizumab Response (TCIA-OBR) dataset and a in-house Predictovar dataset, a retrospective cohort of patients from the Karolinska University Hospital in Stockholm, Sweden. We achieve high prediction test accuracies on treatment resistance of 0.83 and 0.78 on TCIA-OBR and Predictovar dataset, respectively. We further delve into the interpretability of the proposed model by visualizing the feature interactions from patient clinical variables such as patient age and patch-level contributions to the model's predictions. Our proposed approach is a step toward understanding the factors that influence treatment effectiveness in ovarian cancer patients.

Speaker(s):
Emily Nguyen, PhD student
University of Southern California

Author(s):
Zijun Cui, PhD - USC; Joseph Carlson, PhD, MD - USC; Yan Liu, PhD - USC;

Leveraging Machine Learning Explanations to Assess the Appropriateness of Using Demographic Information During Schizophrenia Onset Prediction

Poster Number: P190
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Deep Learning, Fairness and elimination of bias, Machine Learning
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

We create a temporal model that predicts schizophrenia onset for psychosis patients. We construct one variant of this model that uses demographic features and one that does not. Using a model-agnostic feature explanation method, we find that the demographic-aware model explicitly uses demographic information and hypothesize that the demographic-unaware model is implicitly using demographic information through proxy variables. Through this work, we demonstrate the value of feature interpretation for robustly assessing equity in clinical models.

Speaker(s):
Aparajita Kashyap, BA
Columbia University Department of Biomedical Informatics

Author(s):
Aparajita Kashyap, BA - Columbia University Department of Biomedical Informatics; Noemie Elhadad, PhD - Columbia University; Steven Kushner, MD - Columbia University; Shalmali Joshi, PhD - Columbia University;

Augmenting Autism Support: Optimizing the Use of Large Language Models Through Effective Prompt Chaining to Enhance Communication of Therapeutic Guidelines

Poster Number: P191
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Delivering Health Information and Knowledge to the Public, Large Language Models (LLMs), Information Extraction, Clinical Guidelines, Information Retrieval, Documentation Burden
Primary Track: Applications
Programmatic Theme: Clinical Informatics

The complexity of autism, coupled with clinicians' limited time to provide personalized advice for each child, complicates responding to caregivers' frequent inquiries through traditional communication methods. To streamline the process of responding to caregivers' questions, we introduce a Large Language Model (LLM) powered Retrieval Augmented Generation (RAG) with a three-stage prompt chaining process for the generation of autism treatment guidelines. Initial results indicated a preference among clinicians for our RAG model over a general LLM.

Speaker(s):
Deshan Wattegama, MS
University of Missouri

Author(s):
Deshan Wattegama, MS - University of Missouri; Benjamin Black, MD - University of Missouri; Elly Ranum, MD - University of Missouri; Chi-Ren Shyu, PhD, FACMI, FAMIA - University of Missouri-Columbia;

Mixed Methods Assessment of the Influence of Demographics on Medical Advice of ChatGPT

Poster Number: P192
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Diagnostic Systems, Diversity, Equity, Inclusion, Accessibility, and Health Equity, Machine Learning, Fairness and Elimination of Bias, Delivering Health Information and Knowledge to the Public
Primary Track: Applications
Programmatic Theme: Consumer Health Informatics

This study investigates the impact of demographic data on diagnostic consistency between ChatGPT and WebMD symptom checker. Analyzing 540 prompts with varied demographics, ChatGPT demonstrated 91% diagnostic match with WebMD. ChatGPT's urgent care recommendations and demographic tailoring were presented significantly more to 75-year-olds versus 25-year-olds (p<0.01), but were not statistically different among race/ethnicity and sex groups. Readability of ChatGPT's responses exceeded recommended levels, signaling potential health literacy implications.

Speaker(s):
Katerina Andreadis, MS
NYU Grossman School of Medicine

Author(s):
Katerina Andreadis, MS - NYU Grossman School of Medicine; Devon Newman, - - Brown University; Chelsea Twan, MS - NYU Grossman School of Medicine; Amelia Shunk, MMCi - NYU Grossman School of Medicine; Devin Mann, MD - NYU Grossman School of Medicine; Elizabeth Stevens, PhD, MPH - NYU Grossman School of Medicine;

Design and Implementation of an Informatics-Enabled Framework to Improve Race, Ethnicity, and Preferred Language Data Capture in Patients of a Value-Based Primary Care Network

Poster Number: P193
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Diversity, Equity, Inclusion, Accessibility, and Health Equity, Informatics Implementation, Change Management
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Inequalities in healthcare status for historically marginalized and vulnerable populations will persist if systematic collection of data on patient race, ethnicity, and language (REaL) is not prioritized in healthcare settings. After its implementation, a technology-based solution with electronic medical record interoperability improved REaL data capture of older adult patients by 357% for a value-based primary care network across the United States.

Speaker(s):
Kelly Jean Craig, PhD
CVS Health

Author(s):
Kathryn Harris, BA - Oak Street Health, CVS Health; Surabhi Bhatt, MS - Oak Street Health, CVS Health Corporation; Kelly Craig, PhD - CVS Health; Amanda Zaleski, PhD - Aetna; Ali Khan, MD - Oak Street Health, CVS Health;

Old Habits Die Hard – Doctors’ Experiences with Laying Off Free-Text Documentation in the Face of a New, Structured Documentation Regime

Poster Number: P194
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Documentation Burden, Interoperability and Health Information Exchange, Information Retrieval, Knowledge Representation and Information Modeling, Evaluation
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

This study examines the impact of technological and social factors on medical doctors' documentation practices after the introduction of a new EHR (Epic). The study employed qualitative research methods, including participant observations and individual interviews with medical doctors from the largest hospital in Central Norway. We have identified factors that influence doctors’ documentation methods and potential strategies that can promote standardized documentation while ensuring that it aligns with doctors' workflow and maintains their satisfaction.

Speaker(s):
Olga Golburean, MS
NTNU: Norwegian University of Science and Technology

Author(s):
Olga Golburean, MS - NTNU: Norwegian University of Science and Technology; Arild Faxvaag, PhD - NTNU: Norwegian University of Science and Technology; Rune Pedersen, PhD - Norwegian Centre for E-health Research; Line Melby, PhD - NTNU: Norwegian University of Science and Technology;

Exploring Nurses' Perceptions of Documentation Burden in the Electronic Health Record (EHR): A Cross-Sectional Study

Poster Number: P195
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Documentation Burden, Surveys and Needs Analysis, Legal, Ethical, Social and Regulatory Issues
Primary Track: Foundations

This cross-sectional survey examines nurses' perceptions of EHRs documentation burden. The preliminary findings show that the participants identified Flowsheets as the most burdensome documentation activity among ten selected EHR activities. Within the EHR Flowsheet activities, Assessments and Daily Care and Safety flowsheets emerge as the most burdensome. Participants also recommended removal of routinely redundant and double charted information and prefer charting by exception over ‘within defined limits’ documentation policies.

Speaker(s):
Rosie Mugoya, BSN
Goldfarb School of Nursing

Rosie Mugoya, Bsn
Goldfarb School of Nursing and Washington University of St. Louis

Author(s):
Po-Yin Yen, PhD, RN, FAMIA, FAAN - Washington University in St. Louis; Hao Fan, MBBS - Washington University School of Medicine; Sarah Rossetti, RN, PhD - Columbia University Department of Biomedical Informatics; Jennifer Thate, PhD - Siena College;

Human-Centered Design and Development of a Novel Application to Semi-Automate Generation of Cancer Treatment Summaries and Survivorship Care Plans using Electronic Health Record Data

Poster Number: P196
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Documentation Burden, User-centered Design Methods, Administrative Systems, Data Mining, Data Transformation/ETL
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Survivorship care plans (SCPs) are clinical documents written by oncologists and provided to patients that summarize patient-specific cancer treatments, treatment-related health risks, and surveillance recommendations for oncology and hematopoietic stem cell transplant (HSCT) survivors. The manual authorship of SCPs is error-prone and time consuming for healthcare providers (HCPs), with each document requiring up to 12 hours to write, contributing to documentation burden. Computation-supported SCP generation can reduce both provider burden and improve survivorship care, providing cancer survivors and their caregivers with timely access to critically important treatment details, tailored health recommendations and referrals required for personal health management. It is feasible to extract and transform EHR data to support computation-supported SCP generation, which may reduce oncologist documentation burden and enhance health maintenance by cancer survivors. We have developed a proposed user interface for SCPs based on human-centered design methodology and successfully implemented a data pipeline that captures diagnoses and exposure to chemotherapy agents.

Speaker(s):
Andrew Hornback, MS
Georgia Institute of Technology

Author(s):
Areeba Abid, BS - Emory University School of Medicine; Rebecca Lewis, MPH - Children's Hospital of Atlanta; Yuanda Zhu; Benoit Marteau; Benjamin Lovelace, BS - Children's Hospital of Atlanta; Paulette Djachechi, MS - Children's Hospital of Atlanta; James McDaniel, ACM SIG-CHI - Children's Healthcare of Atlanta; May D. Wang, PhD - Georgia Institute of Technology; Naveen Muthu, MD - Children's Healthcare of Atlanta; Karen Effinger, MD, MS - Children's Hospital of Atlanta; Wayne Liang, MD MS FAMIA - Children's Healthcare of Atlanta & Emory University;

Examining the Impact of Gender Imbalance in Training Data on Biases in Deep Learning Models: An Evaluation with Out-of-Distribution Data

Poster Number: P197
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Fairness and Elimination of Bias, Deep Learning, Imaging Informatics, Racial Disparities
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

This study investigates biases in AI models used for predicting findings from chest radiographs, focusing on gender bias stemming from training data. Models trained with varying gender proportions were evaluated on internal and external datasets. Results show generalization gap was not correlated with gender imbalance, with wider generalization gaps for males. Subgroup analysis suggests interactions between gender, race, and bias. Findings underscore the importance of diverse training data to alleviate bias in clinical AI applications.

Speaker(s):
Frank Li, PhD
Emory University

Author(s):
Theo Dapamede, MD, PhD - Emory University; Bardia Khosravi, MD - Mayo Clinic; Saptarshi Purkayastha, PhD - Indiana University, Luddy School of Informatics, Computing and Engineering; Hari Trivedi, MD - Emory University; Judy Gichoya, MD - Emory University;

Anomaly Detection for Medical Data Synthesis Evaluation: A Case Study on Opioid Misuse

Poster Number: P198
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Governance of Artificial Intelligence, Machine Learning, Delivering Health Information and Knowledge to the Public, Population Health, Data Sharing, Privacy and Security, Deep Learning
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Synthetic medical data has been shown to bridge the gap between medical data security, privacy, and data sharing. Over the past years, a massive of data synthesis approaches have been developed driven by rapid developments of generative AI, especially with such as generative adversarial networks (GANs) models. However, little attention has been paid to the quality evaluation of these newly developed models. This paper takes advantage of anomaly detection methods, which do not require prior knowledge or groundtruth information from the dataset, to evaluate the quality of synthetic data generation. The proposed framework enables unsupervised evaluation of anomaly transfer efficiency between real and synthetic data. Using opioid misuse data as a case study, we generated a corresponding synthetic dataset and conducted a comprehensive investigation into the performance of anomaly detection transfer between the real and synthetic datasets.

Speaker(s):
Yili Zhang, PhD
Georgetown University

Author(s):
Yili Zhang, PhD - Georgetown University; Bai Xue, PhD - University of Maryland, Baltimore County; Jia Li Dong, MS - Georgetown University; Yanbao Xiong, MS - Medstar Health; Samir Gupta, PhD - Georgetown University; Maarten Van Segbroeck, PhD - Gretel.ai; Nawar Shara, PhD; Peter McGarvey, PhD - Georgetown University Medical Center;

Implementing Quality Management Systems to Close the AI Translation Gap and Facilitate Safe, Ethical, and Effective Health AI Solutions

Poster Number: P199
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Governance of Artificial Intelligence, Patient Safety, Healthcare Quality
Primary Track: Foundations
Programmatic Theme: Clinical Informatics

The integration of artificial intelligence (AI) into healthcare settings presents transformative potential for improving patient care; however, the translation of AI research and development into clinical practice is hindered by various challenges, including regulatory complexities and the need for robust quality assurance and control measures. This work proposes the adoption of Quality Management System (QMS) frameworks used in regulated industries, such as drug and device development and manufacturing, to ensure the trustworthiness of AI technologies in healthcare settings. This work published in NPJ Digital Medicine aimed to establish a comprehensive framework that accelerates the safe, ethical, and effective delivery of AI/ML solutions in day-to-day patient care, thereby closing the translation gap between AI research and development and clinical practice. QMS elements such as issues and safety surveillance, appropriate policies and procedures, and governance structures can be leveraged to ensure the development of high-quality products, as well as their control and ethical use.

Speaker(s):
Tracey Brereton, Masters
Mayo Clinic

Author(s):
Michael Pencina, PhD - Duke University School of Medicine; John Halamka, MD, MS - Mayo Clinic; Shauna Overgaard, PhD - Mayo Clinic; Tracey Brereton, Masters - Mayo Clinic; David Vidal, JD - Mayo Clinic; Megan Graham, MS - Graham Quality Consulting;

A collaborative approach to the development and implementation of a statistically fair predictive model to support timely and equitable access to specialty palliative care in a large integrated healthcare system

Poster Number: P200
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Health Equity, Advanced Disease, Fairness and Elimination of Bias, Healthcare Quality
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

We present a collaborative approach to model building, validation and implementation. This approach engages clinicians and clinical administrators from model conceptualization to implementation to assure that our clinical decission support tool is clinically relevant, algorithmic equitable, practical for and accepted by clinicians and clinical administrators.

Speaker(s):
Claudia Nau, PhD
Kaiser Permanente

Author(s):
Huong Nguyen, PhD RN - Kaiser Permanente; Lori Viveros, MA - Kaiser Permanente; Susan Wang, MD - Kaiser Permanente;

Continuous Quality Improvement in Social Needs Screening: Evaluation of an Intervention in Bariatric Specialty Care

Poster Number: P201
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Health Equity, Internal Medicine or Medical Subspecialty, Qualitative Methods
Primary Track: Applications
Programmatic Theme: Clinical Informatics

There has been relatively limited investigation into social needs screening and referral programs in specialty care settings. We qualitatively evaluated the implementation of a screening and referral program in a bariatric surgery clinic. Interviews revealed themes related to screening-referral program structure, including variation in patient eligibility for resources and the need for enhanced staff capacity and data system integration. In addition, process-related themes including patient hesitation toward referrals, variability in screening pathways, and uncertainty surrounding communication practices.

Speaker(s):
Bradley Iott, MPH, PhD
University of Michigan

Author(s):
Claire Chang, BA - University of Michigan; Samantha Cooley, MSW - Kansas City University; Jordan Green, MPH - University of Michigan; Dilhara Muthukuda, MPH - University of Michigan; Renuka Tipirneni, MD MSc - University of Michigan;

Developing Quality Metrics for Home-based Teleconsultation in Secondary Stroke Prevention: A Modified Delphi Study

Poster Number: P202
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Telemedicine, Internal Medicine or Medical Subspecialty
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

Background:
Since COVID-19, many Ontario Stroke Prevention Clinics (SPCs) have adopted some level of home-based teleconsultation. However, no quality indicators (QIs) measure the quality of home-based teleconsultation.

Objective:
This study seeks SPC staff’s consensus to rate a set of QIs that can be used to measure the quality of home-based teleconsultation.

Method:
The three-phase Delphi study consisted of a literature preparation followed by one QI-generating focus group meeting, and two subsequent survey rounds were conducted. The SPC staff members in Ontario with teleconsultation experience were invited. The literature review and focus group generated a list of QIs. The panellist rated the proposed QIs in two rounds to assess their clinical relevance and patient orientation. Group consensus was above the Average Percent of Majority Opinion (APMO) cut-off rate.

Results:
Four staff from three SPCs participated in the focus group meeting. Thirteen staff from 13 SPCs responded to the round one survey with a response rate of 31.7% (13/41). The literature review and focus group data synthesis generated 15 QIs. Eight QIs were above the APMO after the first round. The round two survey had a response rate of 92.3% (12/13). The modified QI reached 100% agreement. One of the two new QIs from round one was above the APMO. A total of nine QIs achieved the group consensus.

Conclusions:
The list of QIs could guide the decision-making that supports the utilization of home-based teleconsultation, identifies variations in different models of care and enhances patient safety while using home-based teleconsultation.

Speaker(s):
Guangxia Meng, Master in nursing
University of Waterloo/Southlake Regional Health Center

Author(s):

Effect of secure messaging on EHR time and attention switching

Poster Number: P203
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Healthcare Quality, Workflow, Internal Medicine or Medical Subspecialty, Human-computer Interaction
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This study examined the association between secure messaging use, clinician time spent in the electronic health record (EHR), and attention switching. EHR log data from inpatient providers within a large medical system were used. Secure messaging volume was found to significantly increase EHR time and attention switching, with effects most pronounced among trainee clinicians. These results suggest that secure messaging has a meaningful impact on clinician workflows in the EHR.

Speaker(s):
Daphne Lew, PhD, MPH
Washington University School of Medicine

Author(s):
Daphne Lew, PhD, MPH - Washington University School of Medicine; Laura Baratta - Washington University School of Medicine in St. Louis; Linlin Xia, Ph.D. - Washington University; Elise Eiden, MS - Washington University School of Medicine; Christine Sinsky, MD - American Medical Association; Thomas Kannampallil, PhD - Washington University School of Medicine; Sunny Lou, MD, PhD - Washington University, St. Louis;