Development and Evaluation of a Data Science Workshop for Residents, Researchers, and Students in Psychiatry and Psychology

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

Abstract Keywords: Biomedical Informatics and Data Science Workforce Education, Fairness and Disparity Research in Health Informatics, Natural Language Processing, Education and Training
Primary Track: Clinical Research Informatics
Programmatic Theme: Health Data Science and Artificial Intelligence Innovation: From Single-Center to Multi-Site

As part of our Penn Innovation in Suicide Prevention Implementation Research (INSPIRE) Center, we developed and implemented a three-hour workshop to educate trainees from the psychiatry and psychology departments on using AI as a tool for suicide prevention research with a health equity focus. Evaluation results showed that the workshop enhanced participants’ knowledge and coding abilities.
Speaker(s):
Hayoung Donnelly, Ph.D.
University of Pennsylvania

Author(s):
Hayoung Donnelly, Ph.D. - University of Pennsylvania; David Mandell, ScD - University of Pennsylvania; Sy Hwang, MS - University of Pennsylvania; Emily Schriver, MS - University of Pennsylvania; Ugurcan Vurgun, PhD - University of Pennsylvania; Graydon Neill, BS - University of Pennsylvania; Esha Patel, BS - University of Pennsylvania; Megan Reilly, MPH - University of Pennsylvania; Michael Steinberg, MA - University of Pennsylvania; Amber Calloway, PhD - University of Pennsylvania; Robert Gallop, PhD - University of Pennsylvania; Gregory Brown, PhD - University of Pennsylvania; Danielle Mowery, PhD, MS, MS, FAMIA - University of Pennsylvania;

Development and Evaluation of Patient-Facing Prostate Cancer Decision-Aiding App Featuring Predictive Models

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

Abstract Keywords: Biomedical Informatics and Data Science Workforce Education, Health Literacy Issues and Solutions, Clinical Decision Support for Translational/Data Science Interventions, Health Information and Biomedical Data Dissemination Strategies, Machine Learning, Generative AI, and Predictive Modeling
Primary Track: Clinical Research Informatics
Programmatic Theme: Translational Bioinformatics Using Multi-Modal Patient Data and AI

Prostate cancer is a leading cause of cancer-related deaths. This study developed a web-based app to support personalized treatment decisions by integrating predictive models for EPIC domain scores and specific outcome probabilities for up to 10 years. Using an iterative design process, we improved system usability and enhanced comprehension of treatment outcomes, demonstrated in a pilot study. Future work includes mobile integration and EHR compatibility. Available at: https://cqs.app.vumc.org/shiny/ProstateCancer/.
Speaker(s):
Ashley Kim, BA
Williams College

Author(s):
Ashley Kim, BA - Williams College; Bashir Al Hussein Al Awamlh, MD, MPH - Weill Cornell Medicine; Adam Wright, PhD - Vanderbilt University Medical Center; Allison McCoy, PhD, ACHIP, FACMI, FAMIA - Vanderbilt University Medical Center;

LOINC Implementation Approaches in Academic Medical Research Centers – Early Results from a Survey of CTSA Sites

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

Abstract Keywords: Clinical and Research Data Collection, Curation, Preservation, or Sharing, Data/System Integration, Standardization and Interoperability, Data Standards, Data Transformation/ETL, Reproducible Research Methods and Tools, Secondary Use of EHR Data, Sustainable Research Data Infrastructure
Primary Track: Clinical Research Informatics
Programmatic Theme: Emerging Best Practices for Clinical Research Informatics Operations

Laboratory test results are important for research, yet ubiquitous use of standardized codes (Logical Observation Identifiers Names and Codes, LOINC) remains elusive. This poster will present early results from a survey of CTSA sites on the level and approach of LOINC implementation. Among the 64 CTSA-supported hubs, 19 (30%) responded to the survey. Early results illustrate extreme variation in funding, approaches and scope of LOINC implementation across 19 CTSA institutions.
Speaker(s):
David Hanauer, MD
University of Michigan

Author(s):
Boyd Knosp, MS, FAMIA - University of Iowa; Thomas Campion, PhD - Weill Cornell Medicine;

What Types of Clinical Notes are Used by Researchers?

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

Abstract Keywords: Clinical and Research Data Collection, Curation, Preservation, or Sharing, Informatics Research/Biomedical Informatics Research Methods, Secondary Use of EHR Data
Primary Track: Clinical Research Informatics
Programmatic Theme: Emerging Best Practices for Clinical Research Informatics Operations

Electronic health record notes are critical for clinical research. We analyzed the audit logs from the free text EMERSE search engine to prioritize notes for de-identification. In this period, 527 research projects accessed 529,766 notes, compared to 332,253 notes for non-research use. Key note types included Progress Notes (22%) and Encounter Summaries (20%). Researchers accessed older notes than non-research users. This study underscores the importance of broad access to notes for research.
Speaker(s):
David Hanauer, MD
University of Michigan

Author(s):
David Hanauer, MD - University of Michigan; Lisa Ferguson, MS - University of Michigan; Kellen McClain - University of Michigan; Guan Wang, MS - University of Michigan;

Optimizing Data Collection: Assessing EHR-to-EDC Data Transfer Potential Across Structured and Unstructured Data

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

Abstract Keywords: Clinical and Research Data Collection, Curation, Preservation, or Sharing, Clinical Trials Innovations, FHIR
Primary Track: Clinical Research Informatics
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

This study examined the potential for EHR-to-EDC data transfer in 5 prospective Phase I-III oncology clinical trials, focusing on both structured and unstructured data. The analysis revealed that over 80% of trial data, across structured and unstructured data categories, is transferable for multiple study phases and disease types. The findings underscore the potential capability of EHR-to-EDC technology to transfer both structured and unstructured data to streamline clinical trial processes and improve efficiency.
Speaker(s):
Miles Deneris, M.S.
Flatiron Health

Author(s):
Paul Salcuni, MPH - Flatiron Health; Nadir Ammour, DDS, MBA - Sanofi; Manon Cariou, PharmD - Sanofi; Addison Shelley, BAS - Flatiron Health; Miles Deneris, MS - Flatiron Health; Jessica Herrick, BA, BSRS - Sanofi; Dawn Snow, BA - Sanofi; Sabri Miled, Eng - Sanofi; Veronique Berthou, MSc - Sanofi; Lindsay Bramwell, MSN, RN - Flatiron Health; Natalie Salituro, BS - Flatiron Health; Ivy Altomare, MD - Flatiron Health; Anosheh Afghahi, MD, MPH - Flatiron Health;

Automated Extraction of Discharge Medications Using a Local Large Language Model: A Pilot Study

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

Abstract Keywords: Clinical and Research Data Collection, Curation, Preservation, or Sharing, Data Mining and Knowledge Discovery, Machine Learning, Generative AI, and Predictive Modeling
Primary Track: Clinical Research Informatics
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

This study evaluates the effectiveness of a local Large Language Model (LLaMA 3.1 8B) in extracting discharge medications, review of systems, and past medical history from clinical notes. We compared LLM extractions with manual extractions using 20 discharge summaries from the MIMIC-IV dataset. The LLM achieved high precision for discharge medications, demonstrating the potential of local LLMs to automate chart reviews while addressing privacy concerns.
Speaker(s):
Seung Wook Lee, Doctor of Medicine
MetroWest Medical Center

Author(s):
Seung Wook Lee, Doctor of Medicine - MetroWest Medical Center;

Mixed Feelings About My Voice: Analyzing Errors in a Voice Disorders Screening Model using the Biopsychosocial Framework

Poster Number: P07
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Clinical Decision Support for Translational/Data Science Interventions, Biomarker Discovery and Development, Measuring Outcomes, Machine Learning, Generative AI, and Predictive Modeling
Primary Track: Clinical Research Informatics
Programmatic Theme: Health Data Science and Artificial Intelligence Innovation: From Single-Center to Multi-Site

Voice disorder diagnosis considers both vocal symptoms and their psychosocial impact. We analyzed an XGBoost screening model's errors using the Voice Handicap Index-10, assessing physical, emotional, and functional impacts of voice symptoms. False positives showed higher physical and functional scores than false negatives. This suggests acoustic-based screening tools can identify patients experiencing psychosocial effects from voice symptoms, even if they're not receiving care, highlighting the importance of patient-reported outcomes in assessment.
Speaker(s):
Ruiqing Fan, MA
University of Houston

Author(s):
Ruiqing Fan, MA - University of Houston; Wendy Humphreys, N/A - University of Rochester; Preethi Mathari, PharmD - Michigan Technological University; Zachary Abrams, PhD - Institute for Informatics at Washington University School of Medicine in St. Louis;

Formative Evaluation of a Genomic Clinical Decision Support Tool in Patient Interactions for Alport Syndrome and Autosomal Dominant Polycystic Kidney Disease

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

Abstract Keywords: Clinical Decision Support for Translational/Data Science Interventions, Clinical Genomics/Omics and Interventions Based on Omics Data, Knowledge Representation, Management, or Engineering
Working Group: Clinical Decision Support Working Group
Primary Track: Clinical Research Informatics
Programmatic Theme: Integrating Clinical Research and Clinical Care Workflows

Objective: This study focuses on conducting a formative evaluation to validate the effectiveness and user acceptance of the gCDS tool, integrating it within clinical workflows to simulate real-time genetic data utilization.

Methods: The formative evaluation of the gCDS tool involved multiple iterative cycles with clinician feedback using detailed case vignettes representative of specific genetic conditions. Clinicians engaged with the gCDS tool in a simulated environment, providing insights on usability, aesthetics, and functionality. Participants completed a post session interview and survey of the SDM process. This included the tool usability and workload National Aeronautics and Space Administration (NASA) Task Load Index, Unified Theory of Acceptance and Use of Technology (UTAUT), Perceived Behavioral Control (PBC) scale, and a modified System Usability Scale (SUS). They will also be interviewed after the session to obtain perceptions on the gCDS tool and resources.

Results: Participants are expected to complete post-session interviews and surveys to evaluate the genetic Clinical Decision Support (gCDS) tool, focusing on its usability and integration into clinical workflows. Although the formative evaluation has not yet been completed, feedback from these evaluations will be crucial in refining the tool’s design to enhance ease of use.

Conclusions: While results are pending, the formative evaluation and subsequent feedback are expected to significantly enhance the utility and usability of the gCDS tool, ultimately aiming to improve decision-making processes in nephrology. The findings from this study will provide valuable insights into the integration of genetic information into clinical practice, supporting broader applications across various medical specialties
Speaker(s):
Darren Johnson
Geisinger Medical Center

Author(s):
Marc Williams, MD - Marc S. Williams; Katrina Romagnoli, PhD, MS, MLIS - Geisinger; Alexander Chang, MD - Geisinger Medical Center; Heather Ramey, MA - Geisinger Medical Center;

A Patient-Facing FHIR Application for Improving Pragmatic Clinical Trial Efficiency and Validity

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

Abstract Keywords: Clinical Trials Innovations, FHIR, Patient-centered Research and Care
Primary Track: Clinical Research Informatics
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

By embedding trials in clinical workflows, researchers can rely on extant data from electronic health records (EHRs), rather than prospective data collection. However, the limitations of EHR data quality, including fragmentation across healthcare providers and systems, undercut the validity of research conducted with these data. FHIR offers a solution through the extraction of data from multiple EHRs. We will leverage a patient-facing FHIR application to integrate clinical data from multiple sources, enabling two clinical trials.
Speaker(s):
Emma Young, MS
Oregon Health & Science University

Author(s):
Nicole Weiskopf, PhD - Oregon Health & Science University; Emma Young, MS - Oregon Health & Science University; Michelle Bobo, BA, MS - Oregon Health & Science University; Matthew Storer, BS - Oregon Health & Science University; Patricia Dykes, PhD, MA, RN, FAAN, FACMI - Harvard Medical School; Lipika Samal, MD - Brigham and Women's Hospital; David Dorr, MD - Oregon Health & Science University;

traumaScanner: A text-based algorithm for identifying patients with post-traumatic hemorrhage from radiology text reports

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

Abstract Keywords: Cohort Discovery, Secondary Use of EHR Data, Reproducible Research Methods and Tools
Primary Track: Clinical Research Informatics
Programmatic Theme: Translational Bioinformatics Using Multi-Modal Patient Data and AI

There is a need to curate retrospective cohorts of patients with post-traumatic hemorrhage after traumatic brain injury (TBI) to support clinical risk model development. We devised traumaScanner, a text-based algorithm, that is able to identify patients with post-traumatic hemorrhage from radiology text reports.
Speaker(s):
Meghan Hutch, PhD
University of Chicago

Author(s):
Meghan Hutch, PhD - University of Chicago; Yuan Luo, PhD - Northwestern University; Andrew Naidech, MD MSPH - Northwestern University Feinberg School of Medicine;

All of Us Research Program Participants with Multiple Chronic Conditions

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

Abstract Keywords: Cohort Discovery, Informatics Research/Biomedical Informatics Research Methods, Patient-centered Research and Care, Reproducible Research Methods and Tools
Primary Track: Clinical Research Informatics
Programmatic Theme: Digital Health Technologies for Patient Research

This study estimated the prevalence of MCC for adult All of Us Research Program participants. A cross-sectional analysis was performed to estimate the number of participants with MCC. Relevant diagnostic codes and the corresponding date of diagnosis were extracted for 58 conditions. Most participants were diagnosed with MCC; nearly half with 6+ conditions. The All of Us Research Program is vaulable for analyzing chronic condition prevalence and there is notable prevalence of MCC throughout adulthood.
Speaker(s):
Carolina Gustafson, PhD
University of Pittsburgh

Author(s):
Xintong Li, Bachelor of Science - University of Rochester; Caitlin Dreisbach, PhD, RN - University of Rochester; Carolina Gustafson, PhD - University of Pittsburgh; Komal Murali, PhD - Rory Meyers College of Nursing, New York University; Theresa Koleck, PhD, BSN, RN - University of Pittsburgh;

Breaking Down Silos: Seamless Integration of mHealth and Electronic Health Records (EHRs)

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

Abstract Keywords: Data/System Integration, Standardization and Interoperability, Implementation Science and Deployment, Recruitment Technologies, Data Integration
Primary Track: Clinical Research Informatics
Programmatic Theme: Integrating Clinical Research and Clinical Care Workflows

This study examines the potential of Xealth's intermediary service to improve interoperability between mobile health (mHealth) systems and electronic health records (EHRs). A pilot implementation at the UPMC Adult Spina Bifida Clinic integrated an mHealth system, iMHere 2.0, with EpicCare EHR using Xealth. Results showed successful integration with high conversion rates for app delivery.
Speaker(s):
I Made Agus Setiawan, PhD
University of Pittsburgh

Author(s):
I Made Agus Setiawan, PhD - University of Pittsburgh; Haomin Hu, Master of Science - University of Pittsburgh School of Health and Rehabilitation Sciences; Andi Saptono, PhD - University of Pittsburgh School of Health and Rehabilitation Services; Bambang Parmanto, PhD - University of Pittsburgh; Brad E Dicianno, MD - University of Pittsburgh;

Racial Disparity in Obstructive Sleep Apnea Diagnosis and Heart Failure Development Among Middle-Aged Males: Insights from the All of Us Dataset

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

Abstract Keywords: Data-Driven Research and Discovery, Ethical, Legal, and Social Issues, Fairness and Disparity Research in Health Informatics, Secondary Use of EHR Data
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

Obstructive Sleep Apnea (OSA) is a common but severely underdiagnosed condition, especially among black males, as indicated by prior research. In the United States, OSA is estimated to affect approximately 39 million U.S. adults, while only 6 million cases have been diagnosed. This underdiagnosis is exacerbated by the fact that OSA symptoms occur during sleep and require polysomnography, an expensive and logistically challenging procedure, for confirmation. OSA has been linked to heart failure (HF) through mechanisms such as intermittent hypoxia and increased cardiac workload. Shared risk factors for both conditions include aging, hypertension, obesity, and Type 2 Diabetes Mellitus (T2DM). However, racial disparities in the diagnosis and management of OSA, particularly among black males, have yet to be thoroughly explored. In this study, we address these gaps by examining the relationship between OSA diagnosis and heart failure development using a large, diverse cohort from the All of Us database. Our primary objective was to uncover disparities in diagnosis and their impact on the development of HF in black males compared to white males.
Speaker(s):
Jason Williams, BS
University of Pittsburgh

Author(s):
Olga Kravchenko, PhD - University of Pittsbugh; Kevin Kindler, MD - UPMC; Jason Williams, Health Informatics - University of Pittsburgh;

Use of a voice dictation application among physicians-in-training

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

Abstract Keywords: Education and Training, Data/System Integration, Standardization and Interoperability, Digital Research Enterprise, Informatics Research/Biomedical Informatics Research Methods, Secondary Use of EHR Data
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

Introduction
Physicians-in-training spend a vast amount of time in a computerized environment, much of which involves typing notes. Voice dictation applications are available in some hospital systems as a means to improve typing efficiency and quality. Kaleida Health (Buffalo, NY) uses DMO (Dragon Medical One) by Nuance Communications. We sought to explore the DMO usage patterns among residents and fellows.

Methods
We extracted DMO adoption data between July 2021 and June 2023. This included data on each user's monthly time spent dictating, monthly lines dictated, monthly average dictation speed in lines per hour, monthly DMO auto-text usage, and the total number of commands given each month.

Results
The study included 414 residents and fellows in 2,591 usage months (on average, 6.26 months of DMO usage per physician, out of 24 months). They exhibited significant variability in their monthly DMO usage, where monthly hours dictated ranged from 0 to 19 (median: 1), lines dictated from 0 to 11,505 (median: 428), average lines per hour from 0 to 1,341 (median: 502), DMO auto-text commands from 0 to 595 (median: 0), and total commands from 0 to 1,061 (median: 2).

Discussion and Conclusions
Physicians-in-training in Kaleida Health used DMO for about a quarter of their training months. Their adoption pattern was defined by DMO as "Adopted" or above most of the time. The skewed metrics may reflect previous experience or clinical site heterogeneity. Almost all the commands used were auto-text, which may reflect the need for better DMO training.
Speaker(s):
Samuel Tiosano, MD, MPH
University at Buffalo

Author(s):
Noah Stanco, MD - University at Buffalo Jacobs School of Medicine and Biomedical Sciences; Anneliese Markus, BSc - University at Buffalo; Peter Elkin, MD, MACP, FACMI, FNYAM, FAMIA, FIAHSI - Department of Biomedical Informatics, Jacobs School of Medicine and Biomedical Sciences, University at Buffalo, State University of New York; Michele Lauria, MD - Kaleida Health;

Integration of CIPHER and OHDSI phenotype libraries

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

Abstract Keywords: EHR-based Phenotyping, Knowledge Representation, Management, or Engineering, Phenomics and Phenome-wide Association Studies, Secondary Use of EHR Data
Primary Track: Clinical Research Informatics
Programmatic Theme: Emerging Best Practices for Clinical Research Informatics Operations

Phenotype libraries expedite research using electronic health records by allowing reuse of curated phenotype definitions, and several libraries are publicly available. However, there have been no attempts to integrate libraries to date. We tested the integration of Observational Health Data Sciences and Informatics (OHDSI) phenotype definitions into the Centralized Interactive Phenomics Resource (CIPHER) library by mapping metadata standards. We successfully demonstrated with 25 phenotypes that the use of conceptual standards across libraries enables their integration.
Speaker(s):
Jacqueline Honerlaw, RN, MPH
VA Boston Healthcare System

Author(s):

Workforce and Infrastructure Needs for Reporting Computable Phenotypes in Regulated Research Investigations Using Real-World Data

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

Abstract Keywords: EHR-based Phenotyping, Real-World Evidence and Policy Making, Stakeholder (i.e., patients or community) Engagement
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

Methods for reporting computable phenotypes (CPs) or assessing real-world data (RWD) sufficiency and integrity in regulated research submissions are underdeveloped. This poster identifies outstanding infrastructure and workforce training needs that will assist research sponsors and FDA reviewers in using RWD and CPs, including a platform to register and share CPs and training focused on data management, data transformation, and reporting of RWD. These efforts are expensive and will require collaboration and shared investment.
Speaker(s):
Jennylee Swallow, MS
University of Michigan

Author(s):
Jennylee Swallow, MS - University of Michigan; Rachel Richesson, PhD, MPH, FACMI - University of Michigan Medical School;

Development of a Computable Phenotype for Opioid-induced Adverse Events Using Structured Electronic Health Record (EHR) Data

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

Abstract Keywords: EHR-based Phenotyping, Drug Discovery, Repurposing, and Side-effect Discovery, Secondary Use of EHR Data
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

We developed a computable phenotype to identify opioid-induced adverse events in pediatric patients using structured EHR data. By analyzing inpatient encounters from 2016 to 2023 at Duke University Hospital, we examined patterns of naloxone administration, a key indicator for adverse events such as opioid-induced respiratory depression. The study highlights how different naloxone administration methods—infusion versus bolus—correlate with adverse event severity, providing insights to improve the accuracy of future retrospective opioid research in children.
Speaker(s):
Rushi Tang, Master of Biostatistics (In Progress)
Duke University

Author(s):
Rushi Tang, Master of Biostatistics (In Progress) - Duke University; Benjamin Goldstein, PhD - Duke University; Lisa Einhorn, M.D. - Duke University Hospital; Jillian Hurst, Ph.D. - Duke University; Sydney Reed, M.D. - Duke University Hospital;

Multisystem Inflammatory Syndrome in Children across Clinical Sites

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

Abstract Keywords: EHR-based Phenotyping, Cohort Discovery, Data-Driven Research and Discovery, Measuring Outcomes, Clinical and Research Data Collection, Curation, Preservation, or Sharing
Primary Track: Clinical Research Informatics
Programmatic Theme: Health Data Science and Artificial Intelligence Innovation: From Single-Center to Multi-Site

Multisystem inflammatory syndrome in children (MIS-C) is a severe delayed hyperinflammatory condition observed in a small proportion of children two to six weeks after COVID-19 infection. Developing MIS-C cohorts is difficult due to its rarity. Electronic health record data allows high throughput characterization of patients with MIS-C. We tested ICD code-based identification of patients with MIS-C across clinical sites and characterized multiple MIS-C cohorts. Clinical presentation was similar across the sites, but outcomes differed.
Speaker(s):
Vijeeth Guggilla, BA
Northwestern University

Author(s):
Mike Semanik, MD - University of Wisconsin; Dominic Co, MD, PhD - University of Wisconsin; Brian Nolan, MD - Ann & Robert H. Lurie Children's Hospital of Chicago; Lacey Gleason, MSPH - Northwestern University; Alona Furmanchuk, PhD - Center for Health Information Partnerships, Northwestern University, , Feinberg School of Medicine; Judith Smith, MD, PhD - University of Wisconsin; Theresa Walunas, PhD - Northwestern University;

Visualize Phenotype Differences and Similarities: CIPHER’s Comparison Tool

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

Abstract Keywords: EHR-based Phenotyping, Secondary Use of EHR Data, Public Health Informatics
Primary Track: Clinical Research Informatics
Programmatic Theme: Digital Health Technologies for Patient Research

The emergence of new phenotype libraries has helped to facilitate algorithm dissemination and reuse. However, few resources exist to aid in the comparison of multiple phenotypes within these libraries. We created a tool that enables comparison of multiple phenotype definitions in the CIPHER knowledgebase. By presenting a side-by-side analysis of phenotypes and highlighting similarities and differences in metadata, CIPHER's comparison tool helps users to determine the most appropriate definition to apply for their purposes.
Speaker(s):
Tiffany Sim, MPH
Veterans Affairs

Author(s):

The Development and Provision of Large-Scale De-identified Research Database in OMOP Common Data Model

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

Abstract Keywords: Enterprise Data Warehouse/Data Lake, Data Security and Privacy, Data Sharing/Interoperability, Data/System Integration, Standardization and Interoperability
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

To facilitate research while protecting patient privacy, the University of Rochester Medical Center developed a de-identified EHR database covering 2.3 million patients, transformed from Epic into the OMOP common data model. The de-identification approach effectively balances privacy concerns with data utility, minimizing re-identification risks, particularly for rare clinical conditions and procedures. Comprehensive clinical data—including demographics, diagnoses, procedures, medications, labs, visits, vitals, social determinants of health, patient reported outcomes, and more—is securely accessible within a HIPAA-compliant enclave. Policies and technical safeguards ensure privacy, and the initiative has received institutional approvals.
Speaker(s):
Jack Chang, PhD, MS
URMC

Author(s):
Jack Chang, PhD, MS - URMC;

Scaling a Patient Portal Integrated Diabetes Application Using FHIR: A Multisite Experience

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

Abstract Keywords: FHIR, Data Sharing/Interoperability, Data/System Integration, Standardization and Interoperability, Patient-centered Research and Care, Health Literacy Issues and Solutions, Implementation Science and Deployment
Primary Track: Clinical Research Informatics
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

Introduction: We built My Diabetes Care, a patient-facing app using SMART on FHIR for cross-EHR interoperability. Methods: We used the Consolidated Framework for Implementation Research to study facilitators and barriers to implementation across two organizations. Results: Key themes include differences in user authentication (intervention), focus on 21st Century Cures Act (outer setting), competing priorities (inner setting), and limited experience with FHIR (individuals, process). Conclusions: Addressing barriers can optimize digital solutions, improving outcomes and reducing disparities.
Speaker(s):
Nicolás Prada-Rey, MA
Brigham and Women's Hospital

Author(s):
Nicolás Prada-Rey, MA - Brigham and Women's Hospital; William Martinez, MD, MS - Vanderbilt University Medical Center; S. Trent Rosenbloom, MD, MPH, FACMI, FAMIA - Vanderbilt University Medical Center Dept of Biomedical Informatics; Richard Fay, MA - Brigham and Women's Hospital; Zhou Yan, MS - Vanderbilt University Medical Center; Brandon Oglesby, MS - Vanderbilt University Medical Center; Amit Battu, MS - Vanderbilt University Medical Center; Matthew Wien, MS - Brigham and Women's Hospital; Frank Chang, MS - Brigham and Women's Hospital; Adam Wright, PhD - Vanderbilt University Medical Center; Lipika Samal, MD - Brigham and Women's Hospital; Jorge Rodriguez, MD - Brigham and Women's Hospital/Harvard Medical School;

Evaluation of Cancer-Related Nutrition Videos on YouTube: Comparing between Long vs. Short-Form Videos

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

Abstract Keywords: Health Information and Biomedical Data Dissemination Strategies, Stakeholder (i.e., patients or community) Engagement, Data-Driven Research and Discovery
Primary Track: Clinical Research Informatics
Programmatic Theme: Digital Health Technologies for Patient Research

Nutrition is vital for cancer patients, yet many report inadequate information, leading them to seek guidance on social media platforms like YouTube. This study compares long-form and short-form YouTube videos on cancer nutrition, focusing on content types, evidence-based information, and user engagement. A total of 100 videos (50 long-form, 50 short-form) were analyzed. Short-form videos had a median of 614.5 views, focusing primarily on food recommendations for cancer prevention and survivorship. Long-form videos had higher engagement, with a median of 85,000 views, and covered broader topics including nutrition education. Both formats were predominantly produced by healthcare providers. These findings emphasize the need for credible, expert-driven content on social media to ensure accurate nutrition guidance for cancer patients. Future research should assess the quality of information across specific cancer types and explore other platforms like TikTok and Instagram for broader dissemination.
Speaker(s):
Youjia Wang, BSN, RN
University of Pittsburgh School of Nursing

Author(s):
Young Ji Lee, PhD - University of Pittsburgh;

Reporting Resuscitation Team Performance: Integrating EHR and Zoll Defibrillator Data

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

Abstract Keywords: Health Information and Biomedical Data Dissemination Strategies, Data Integration, Secondary Use of EHR Data, Implementation Science and Deployment
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

There is often a gap between the perceived quality of cardiac arrest care and actual quality delivered.

We developed a report that combines data from defibrillator accelerometers and electronic health records. This report is emailed daily to participants of each event and includes key indicators of resuscitation quality.

When clinicians receive personal performance metrics, they tend to improve. This suggests that providing such feedback could help bridge the gap between perceived and actual care quality.
Speaker(s):
Cody Couperus, MD
University of Maryland Medical Center

Author(s):
Cody Couperus, MD - University of Maryland Medical Center; Mark Sutherland, MD - University of Maryland Medical Center; Chen Dun, MHS - Johns Hopkins University; Samuel Gurmu, BS - University of Maryland Medical Center; Dan Lemkin, MD - University Of Maryland, School of Medicine; Nicholas Morris, MD - University of Maryland Medical Center;

Process Improvement in Code Blue Alerts

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

Abstract Keywords: Implementation Science and Deployment, Learning Healthcare System, Measuring Outcomes, Data-Driven Research and Discovery
Primary Track: Clinical Research Informatics
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

We aimed to improve hospital adherence with American Heart Association Get With The Guidelines – Resuscitation quality metrics by critically examining the Code Blue activation process. We identified and eliminated manual and duplicative steps, provided physical barriers to reduce accidental activations, and emphasized staff education. After piloting the new process, we were able to identify improvements in staff understanding and a low false activation rate.
Speaker(s):
Joshua Derbort, MD
University of Texas Southwestern

Author(s):
Joshua Derbort, MD - UT Southwestern Medical Center; Brooke Powell, BSN - Parkland Hospital; Patrick Malecha, MHA - Parkland Hospital; Frederick Waheed, MBA - Parkland Hospital; Angel Bates, BSN - Parkland Hospital; Melinda Savage, BSN - Parkland Hospital; Catherine Chen, MD - UT Southwestern Medical Center;

Implementing An Enterprise Clinical Trials Recruitment System: The Indiana University CTSI-TrialX Experience

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

Abstract Keywords: Implementation Science and Deployment, Recruitment Technologies, Sustainable Research Data Infrastructure
Primary Track: Clinical Research Informatics
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

We describe the implementation and use of an enterprise-wide clinical trials awareness and recruitment system at Indiana University CTSI. The system has been deployed for approximately five years and has received approximately 250,000 unique visitors. It has become a central resource for making Indiana CTSI studies available to the public and serves as a state-wide registry for potential research volunteers.
Speaker(s):
Chintan Patel
TrialX Inc.

Author(s):
Titus Schleyer, DMD, PhD, MBA - Regenstrief Institute; Waqas Amin, MD, MSIS - Indiana University; Sharib Ahmad Khan - TrialX; Chintan Patel - TrialX Inc.; Medina Sydykanova, PMP, MSHM - Regenstrief Institute; Brenda Hudson, MA, PMP, CCRP - Indiana Clinical and Translational Science, Indiana University;

Assessing Mobile Phone Capacity with MoDAT: A Simulation-Driven Approach

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

Abstract Keywords: Mobile Health, Wearable Devices and Patient-Generated Health Data, Clinical Decision Support for Translational/Data Science Interventions, Reproducible Research Methods and Tools
Primary Track: Clinical Research Informatics
Programmatic Theme: Digital Health Technologies for Patient Research

Individuals with disabilities often struggle to identify suitable accessibility settings, hindering their access to technology. Clinicians also face challenges recommending appropriate options due to lack of experience and training. To address this, we developed MoDAT, a high-fidelity assessment tool simulating real-world smartphone environments. MoDAT evaluates users' physical and cognitive abilities through a mobile app, assisting clinicians in matching individuals with the appropriate accessibility settings.
Speaker(s):
Firdaus Indradhirmaya, PhD Student
University of Pittsburgh

Author(s):
Firdaus Indradhirmaya, PhD Student - University of Pittsburgh; Qingxiang Zeng, Masters in Science - Hari Lab; Andi Saptono, PhD - University of Pittsburgh School of Health and Rehabilitation Services;

Integrating a Family Caregiver Risk Screening Tool into Clinic: Usability Testing among Family Caregivers of People with Gynecologic Cancer

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

Abstract Keywords: Mobile Health, Wearable Devices and Patient-Generated Health Data, Patient-centered Research and Care, Clinical and Research Data Collection, Curation, Preservation, or Sharing, Stakeholder (i.e., patients or community) Engagement
Primary Track: Clinical Research Informatics
Programmatic Theme: Digital Health Technologies for Patient Research

Screening to identify potential risk factors may help family caregivers prevent and mitigate poor outcomes, yet few screening tools are available. We trialed a developed risk-screening tool in a gynecological cancer clinic to test its usability and explore the potential of officially adapting it to in-clinic caregiver support. Our findings suggest that the digital screening tool is usable and well-received by FCGs. Lessons regarding the tool’s iterative improvements and implementations were gathered from user feedback.
Speaker(s):
Haomin Hu, Master of Science
University of Pittsburgh School of Health and Rehabilitation Sciences

Author(s):
Youjia Wang, BSN, RN - University of Pittsburgh School of Nursing; Julie Klinger, Master of Art - University of Pittsburgh; Grace Campbell, Ph.D. - Duquesne University; I Made Agus Setiawan, PhD - University of Pittsburgh; Yong Choi, PhD, MPH - University of Pittsburgh; Bambang Parmanto, Ph.D. - University of Pittsburgh; Heidi Donovan, PhD, RN;

Predicting Clozapine Prescription in Treatment of Schizophrenia: A Comparative Study of Machine Learning Models Using Structured Data and Clinical Notes

Poster Number: P28
Presentation Time: 05:00 PM - 06:30 PM

Abstract Keywords: Natural Language Processing, Machine Learning, Generative AI, and Predictive Modeling, Secondary Use of EHR Data
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

This study integrates natural language processing (NLP) and machine learning to predict clozapine prescriptions for treatment-resistant schizophrenia (TRS) using structured and unstructured electronic health record data. Deep Neural Networks (DNN) and Random Forest showed the highest predictive accuracy, with GatorTron embeddings performing best in identifying TRS patients for clozapine. These findings support the development of a high-throughput computable phenotype to enhance clinical management and decision-making for TRS patients. This study integrates natural language processing (NLP) and machine learning to predict clozapine prescriptions for treatment-resistant schizophrenia (TRS) using structured and unstructured electronic health record data. Deep Neural Networks (DNN) and Random Forest showed the highest predictive accuracy, with GatorTron embeddings performing best in identifying TRS patients for clozapine. These findings support the development of a high-throughput computable phenotype to enhance clinical management and decision-making for TRS patients.
Speaker(s):
Author(s):

A Preliminary Study of Incarceration and Military Service Status Among Individuals Tested for Hepatitis C Virus Infection Using Natural Language Processing

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

Abstract Keywords: Natural Language Processing, Social Determinants of Health, Secondary Use of EHR Data
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

Hepatitis C virus (HCV) infection remains a public health concern in the U.S., with history of incarceration and military service being key risk factors for infection. We used electronic health records and natural language processing to extract information on these risk factors from clinical narratives. The GatorTron-based model achieved the best performance for concept extraction of incarceration history status (precision: 0.8919, recall: 0.9167, F1-score: 0.9041) and military service status (precision: 0.9474, recall: 0.9231, F1-score: 0.9351).
Speaker(s):
Pilar Hernandez-Con, MD, MSCE
University of Florida

Author(s):
Daniel Paredes Pardo, MS - Department of Health Outcomes and Bioinformatics, College of Medicine, University of Florida; Chanakan Jenjai, PharmD - Department of Pharmaceutical Outcomes and Policy, College of Pharmacy, Unversity of Florida; Shunhua Yan, MEd - Department of Pharmaceutical Outcomes and Policy, College of Pharmacy, University of Florida; Ashley Stultz, BS - Department of Pharmaceutical Outcomes and Policy, College of Pharmacy, University of Florida; Debbie Wilson; Khoa Nguyen, Pharm.D - University of Florida; Jenny Lo-Ciganic, PhD - University of Pittsburg; Ashley Norse, MD - University of Florida Health Jacksonville; Faheem Guirgis, MD - College of Medicine, University of Florida; Robert Cook, MD, MPH - Department of Epidemiology, University of Florida; David Nelson, MD - University of Florida; Haesuk Park, PhD - Department of Pharmaceutical Outcomes, College of Medicine, University of Florida; Yonghui Wu, PhD - University of Florida;

Towards a More Comprehensive Medication Ontology in the ENACT Network

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

Abstract Keywords: Ontologies, Cohort Discovery, EHR-based Phenotyping
Primary Track: Clinical Research Informatics
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

As EHR data research gains popularity, investigators have become increasingly reliant on assembling high-quality concept sets. Building medication concept sets for querying and analyses can be a time-consuming task. Medication ontologies are an effective way to organize drugs by including constituents, dosing formats, and NDCs. We are developing an improved drug ontology for cohort identification and analysis. The goal of this ontology is to provide more specific categories and complete class-to-ingredient mapping, while lowering investigator effort and improving the quality of medication-based queries in the ENACT network.
Speaker(s):
Michele Morris, BA
University of Pittsburgh

Author(s):

HemOnc.org: A Growing Hematology/Oncology Resource for Health Care Providers and Translational Researchers

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

Abstract Keywords: Ontologies, Real-World Evidence and Policy Making, Biomedical Informatics and Data Science Workforce Education
Primary Track: Clinical Research Informatics
Programmatic Theme: Integrating Clinical Research and Clinical Care Workflows

HemOnc.org is a comprehensive, dynamic, and freely accessible knowledge base of evidence-based treatments for cancer and blood disorders, developed as a crowdsourced wiki since 2011. Authorized healthcare professionals and trainees can contribute to this resource, which, as of September 6, 2024, includes 5,262 treatment regimens across 191 distinct cancers and blood disorders. The site details individual drug components, sequence, dose, route, and supporting clinical trial information. Portions of HemOnc.org have been transformed into a computable ontology in the Observational Medical Outcomes Partnership (OMOP) format. The HemOnc ontology is available for academic and non-commercial use and has been adopted by the Observational Health Data Sciences and Informatics (OHDSI) program and integrated into the National Cancer Institute Thesaurus. Future goals include expanding existing content, increasing international usership, and building new regimen categories based on real-world evidence (RWE) treatment exposures.
Speaker(s):
Aleenah Mohsin, MBBS
Brown University, Rhode Island Hospital

Author(s):
Aleenah Mohsin, MBBS - Brown University, Rhode Island Hospital; Sandeep Jain, MD; Sanjay Mishra, PhD; Andrew Cowan, MD - Fred Hutchinson Cancer Center, Seattle, WA, USA; Wayne Liang, MD MS FAMIA - Children's Healthcare of Atlanta & Emory University; Jessica Patricoski, PhD student, Computational Biology - Brown University Center for Computational Molecular Biology; Alexander VanHelene, BS - Brown University; Peter Yang, MD - HemOnc.org LLC, Lexington, MA, USA; Jeremy Warner, MD, MS - Brown University;

Assessing the Accuracy of ICD-10 Coding for COVID-19 Infections in Critically Ill Patients

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

Abstract Keywords: Public Health Informatics, Informatics Research/Biomedical Informatics Research Methods, Data/System Integration, Standardization and Interoperability
Primary Track: Clinical Research Informatics
Programmatic Theme: Integrating Clinical Research and Clinical Care Workflows

Our study aims to examine the accuracy of ICD coding for COVID-19 in the critically ill population, which is a sparser area of research. Our results found that ICD coding remains an unreliable method of healthcare surveillance and that there may be more utility in deferring to laboratory testing for COVID-19 surveillance until ICD coding practices improve.
Speaker(s):
Melissa Bak, BA
Northwestern University Feinberg School of Medicine

Author(s):
Melissa Bak, BA - Northwestern University Feinberg School of Medicine; Mengjia Kang - Northwestern University; Luke Rasmussen, MS - Northwestern University; Theresa Walunas, PhD - Northwestern University; Richard Wunderink, MD - Northwestern University; Catherine Gao, MD - Northwestern University;

The MassCPR Federated Data Network for Pathogen Readiness

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

Abstract Keywords: Public Health Informatics, Clinical and Research Data Collection, Curation, Preservation, or Sharing, Real-World Evidence and Policy Making, Data Mining and Knowledge Discovery, Data Security and Privacy
Primary Track: Clinical Research Informatics
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

The Massachusetts Consortium on Pathogen Readiness (MassCPR) is launching a Federated Data Network to uniquely connect hospital electronic health records, public health datasets, and biospecimens. Based on the widely adopted i2b2 and SHRINE software platforms, the network will support both research and public health use cases related to infectious diseases. We expect it to serve as a model that can be extended to other states across the country.
Speaker(s):
Griffin Weber, MD, PhD
Harvard Medical School

Author(s):
Shawn Murphy, MD, Ph.D. - Massachusetts General Hospital;

Utilization of the FDA Adverse Events Reporting System (FAERS) to Identify Spontaneously Reported Cardiac Adverse Event Signals for Natural Products

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

Abstract Keywords: Real-World Evidence and Policy Making, Data-Driven Research and Discovery, Public Health Informatics
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

The use of natural products such as herbal supplements for perceived health benefits has become prevalent in the United States. However, the safety risks associated with these natural products are still uncertain with limited available data. We aim to identify cardiac adverse event signals for natural products using real-world data from spontaneous reports submitted to the publicly available FDA adverse events reporting system.
Speaker(s):
Christina Kazarov, PharmD
University of Pittsburgh School of Pharmacy

Author(s):
Sandra Kane-Gill, PharmD, MSc - University of Pittsburgh School of Pharmacy; Richard Boyce, PhD - University of Pittsburgh; Christina Kazarov, PharmD - University of Pittsburgh School of Pharmacy;

Exploring Reproducibility Issues Related to the Use of Large Language Models in the Clinical Domain

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

Abstract Keywords: Reproducible Research Methods and Tools, Informatics Research/Biomedical Informatics Research Methods, Machine Learning, Generative AI, and Predictive Modeling
Primary Track: Clinical Research Informatics
Programmatic Theme: Harnessing the Power of Large Language Models in Health Data Science

This study explores the reproducibility issues in research involving large language models (LLMs) applied to clinical topics. We reviewed 173 relevant studies from PubMed to quantify and assess methodological and reporting practices. Key findings include a lack of reporting on LLM versions and parameters, and variability in human annotator expertise. These shortcomings could compromise reproducibility. We recommend standardized reporting guidelines to enhance research transparency and reliability in this growing field.
Speaker(s):
David Hanauer, MD
University of Michigan

Author(s):
David Hanauer, MD - University of Michigan; Yunan Chen, PhD - University of California, Irvine; Kai Zheng, PhD - University of California, Irvine;

Creating a Transplant-Specific Comorbidity Index: Validation of Existing Models and Development of New Predictive Tools

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

Abstract Keywords: Secondary Use of EHR Data, Outcomes Research, Clinical Epidemiology, Population Health, Data-Driven Research and Discovery
Primary Track: Clinical Research Informatics
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

This project aims to validate the Charlson and Elixhauser comorbidity indices in the kidney and liver transplant patient population. It also aims to develop a new comorbidity index tailored to this population to better predict mortality. Using Kaplan-Meier plots and log-rank tests, we found that excluding renal and liver failure from the respective transplant populations tended to improve the validation of these indices.
Speaker(s):
Faith Mendoza, Master of Biostatistics
Duke University

Author(s):
Faith Mendoza, Master of Biostatistics - Duke University; Tyler Schappe, MS; Ursula Rogers, BS - Duke University School of Medicine; Roland Matsouaka, PhD - Duke University; Nrupen Bhavsar, PhD - Duke University; Lisa McElroy, MD, MS - Duke University; Samuel Berchuck, PhD - Duke University; Liz Nichols, MB (Master of Biostatistics) - Duke University;

Secure, Seamless, and Unified Slide Repositories: A Distributed Approach for Accessing Medical Data Across Secure Networks

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

Abstract Keywords: Data-Driven Research and Discovery, Data/System Integration, Standardization and Interoperability, Advanced Data Visualization Tools and Techniques, Data Security and Privacy
Primary Track: Data Science/Artificial Intelligence
Programmatic Theme: Digital Health Technologies for Patient Research

Advances in AI for medical research require large-scale, multi-institutional data sharing. To support the Federated digital pathology platform for AD/ADRD research (NIH 1U24NS133945-01), we are developing distributed repositories for whole slide imaging (WSI) data. Using a custom S3-compatible server and secure mesh communications, the platform enables data transfer, localized processing, and federated training across sites. Testing on the FABRIC Testbed addresses challenges like latency and firewalls in a real-world infrastructure.
Speaker(s):
Vaiden Logan, B.S. in Computer Engineering
UKY

Author(s):
Mitchell Klusty, B.S. Computer Science - University of Kentucky; Cody Bumgardner, PhD - University of Kentucky; Ken Calvert, Computer Science - University of Kentucky; Jeffery Talbert, PhD - University of Kentucky;

Discovering Genetic Associations through Brain Imaging Representation Learning with Vision Transformer Autoencoders

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

Abstract Keywords: Data-Driven Research and Discovery, Genotype-phenotype Association Studies (including GWAS), Medical Imaging, Machine Learning, Generative AI, and Predictive Modeling
Primary Track: Data Science/Artificial Intelligence
Programmatic Theme: Health Data Science and Artificial Intelligence Innovation: From Single-Center to Multi-Site

In order to understand how complex brain structures are associated with genetic signals, we utilized a vision-transformer based autoencoder to learn a set of phenotypes from T1 brain MRI and then perform genome-wide association studies (GWAS). Our 384-dimensional vision transformer autoencoder was trained on UK Biobank dataset and achieved a validation reconstruction mean squared error of 0.187. GWAS of these 384 phenotypes revealed 33 new loci compared with a convolutional neural network based autoencoder.
Speaker(s):
Samia Islam, MS
The University of Texas Health Science Center at Houston

Author(s):

Enhancing an alcohol intervention through behavior change techniques and adaptive user personas: Personalized messaging for a UK military cohort

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

Abstract Keywords: Machine Learning, Generative AI, and Predictive Modeling, Public Health Informatics, Real-World Evidence and Policy Making
Primary Track: Data Science/Artificial Intelligence
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

This study explores the efficacy of a personalized AI-driven messaging framework designed to enhance DrinksRation; a brief alcohol intervention for the UK military. By using behavior change and adaptive user personas, the framework tailors message content based on a changing users’ drinking patterns. Preliminary results from a study of 157 participants showed significant increases in app engagement and reductions in alcohol consumption, highlighting the potential of AI to improve outcomes.
Speaker(s):
Daniel Leightley, BSc MSc PhD
King's College London

Author(s):
Grace Williamson, BSc MSc - King's College London; Charlottle Williamson, BSc MSc - King's College London; Iain Marshall, PhD MRCGP - King's College London; Vasa Curcin, PhD - King's College London;

Pathway-anchored dimension reduction and MLOps enable robust micro-cohort transcriptome classifier development

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

Abstract Keywords: Machine Learning, Generative AI, and Predictive Modeling, Transcriptomics, Ontologies
Primary Track: Data Science/Artificial Intelligence
Programmatic Theme: Translational Bioinformatics Using Multi-Modal Patient Data and AI

Abstract: We hypothesized integration of a paired-sample design, single-subject pathway-level analytics, and Machine Learning Operations (MLOps) would enable unbiased classifier development in small, potentially biased cohorts through improvements in signal-to-noise ratio, mechanistically-anchored dimension reduction, and automation, reproducibility, continuous monitoring, as well as reliable and efficient model deployment. We demonstrate effectiveness by developing a small-cohort (n=19) classifier of symptomatic vs asymptomatic viral infections, achieving 92% precision and 90% recall, with 42 biological process features selected among ~5000.
Speaker(s):
Madi Shabanian, phdStudent
University of Utah

Author(s):
Yves Lussier, MD - University of Utah; Nima Pouladi, PhD, MD - University of Utah; Liam S. Wilson, Bachelor - University of Utah; Jianrong Li, Msc - University of Utah;

Use of Machine Learning and Clinical Phenotyping to Predict CVD Risk for Primary Prevention with Data from Electronic Health Records

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

Abstract Keywords: Machine Learning, Generative AI, and Predictive Modeling, Secondary Use of EHR Data, Data-Driven Research and Discovery, EHR-based Phenotyping, Clinical Decision Support for Translational/Data Science Interventions
Primary Track: Data Science/Artificial Intelligence
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

This project aims to improve the performance of CVD risk prediction algorithms by using the optimal ML approaches including supervised learning and neural networks in the EHR data linkage.

Cardiovascular disease (CVD) prediction algorithms have a long history of adoption by physicians in both the UK and US healthcare sectors to predict and calculate the medium- to long-term risk of cardiovascular events. A number of models have been developed, validated, and amended in various population settings and recommended by national clinical guidelines. At the same time, a growing number of risk factors have been incorporated to improve the general performance of these models. The development of clinical decision support systems (CDSSs) using Electronic Health Records (EHR) and Machine/Deep Learning (ML/DL) approaches are valuable opportunities for further refinement for CVD prevention and planning. These three techniques are increasingly used in conjunction with conventional statistical models that are derived from traditional cohort datasets. However, due to a lack of research, the promise of the added value in the predictive performance (discrimination and calibration) of models premised on these techniques remains unclear.
Speaker(s):
Vasa Curcin, PhD
King's College London

Author(s):
Tianyi Liu, PhD student - King's College London; Andrew Krentz, PhD - King's College London; Lei Lu, PhD - King's College London; Vasa Curcin, PhD - King's College London;

A Preliminary Study of Documenting Stigma, Social, and Behavioral Information in Clinical Notes among Patients with HIV

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

Abstract Keywords: Outcomes Research, Clinical Epidemiology, Population Health, Natural Language Processing, Biomedical Informatics and Data Science Workforce Education
Primary Track: Data Science/Artificial Intelligence
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

Stigma, social, and behavioral information remain a significant barrier to HIV prevention and treatment efforts impacting the health of people with HIV (PWH). These issues appear as social ostracism, personal rejection, discrimination, and laws that violate the basic human rights of PWH. However, there is limited understanding of how these exposures affect HIV health outcomes. This study aims to investigate the documenting of stigma, social, and behavioral information among PWH using narrative clinical notes.
Speaker(s):
Ziyi Chen, Master of Science
Department of Health Outcomes and Biomedical Informatics, College of Medicine, University of Florida

Author(s):
Ziyi Chen, Master of Science - Department of Health Outcomes and Biomedical Informatics, College of Medicine, University of Florida; Yiyang Liu, PhD - University of Florida; Mattia Prosperi, PhD, FAMIA - University of Florida; Robert Cook, MD - University of Florida; Krishna Vaddiparti, PhD - University of Florida; Jiang Bian, PhD - University of Florida; Yonghui Wu, PhD - University of Florida;

Bridging The Gap: Innovations in the Classification of Genetic Variants for Clinical Application

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

Abstract Keywords: Biomarker Discovery and Development, Clinical Genomics/Omics and Interventions Based on Omics Data, Genomics/Omic Data Interpretation, Systems Biology and Network Analysis, Clinical Genomics/Omics and Interventions Based on Omics Data, Data Mining and Knowledge Discovery, Informatics Research/Biomedical Informatics Research Methods, Clinical Genomics/Omics and Interventions Based on Omics Data
Working Group: Genomics and Translational Bioinformatics Working Group
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Novel Methods for Variant Detection and Interpretation from Omics Data

The rapid evolution of genomic technologies has propelled the field of genetics into a new era, where the identification of novel genetic variants has become increasingly routine. Particularly in clinical settings, the discovery of these variants holds the potential to revolutionize personalized medicine, offering insights into individual predispositions to diseases, response to medications, and optimal therapeutic strategies. However, this wealth of genomic data presents a significant challenge: the classification of novel variants, especially those of uncertain significance, and their implications for clinical outcomes. As a result, there is a growing gap between the discovery of genetic variants and their application in clinical practice, underscoring the need for innovative approaches to assess their pathogenicity. This paper aims to explore the current landscape of variant classification for novel variants within clinical data, emphasizing the integration of machine learning models.
Our study enriches the variant classification process by incorporating a multidimensional analysis that includes functional predictions, evolutionary conservation scores, structural implications, and indirect clinical evidence. This approach not only enhances our ability to accurately classify variants but also provides a scalable solution to manage the burgeoning volume of genomic data from clinical settings. To facilitate this, we aggregate features from various sources, categorized by knowledge databases such as ClinVar, and computational ranking algorithms. The classification of variant pathogenicity is conducted using a Random Forest algorithm, trained on a model developed from ClinVar data shown as Table 1. The primary features collected for model training are detailed in the accompanying table.
Speaker(s):
Hui Li, Phd
University of Texas Health Science Center at Houston

Author(s):

ASCEND: A Comprehensive Transcriptomic Analysis Tool for Biomarker Identification and Methylation Marker Association in Cancer Predictions

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

Abstract Keywords: Biomarker Discovery and Development, Epigenomics, Transcriptomics, Clinical Genomics/Omics and Interventions Based on Omics Data
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Novel Methods for Variant Detection and Interpretation from Omics Data

Gene expression serves as a fundamental metric in molecular biology, providing insights into cells’ functional outputs and the body’s physiological processes. Epigenetic changes in gene expression often serve as predictive indicators of various cancers, making a robust tool for gene expression and epigenetic cancer analysis imperative. ASCEND is a novel Python-based tool that simultaneously analyzes transcriptomic and DNA methylation data to predict cancer and identify potential biomarker genes and related methylation markers. The algorithm operates in two stages: first predicting cancer presence and identifying biomarker genes using a Multilayer Perceptron Classifier, then determining associated DNA methylation markers through linear regression. Applied to a subset of TCGA Breast Adenocarcinoma data (101 samples), ASCEND demonstrated 87% accuracy in predicting disease presence. It identified five genes strongly associated with the disease: WEE2P1, SUPT20HL1, TBC1D4, DGCR11, and TEX26, with four of these validated by existing medical literature. The tool also pinpointed prominent CpG methylation sites out of 27,577 total sites in the dataset, which can potentially be therapeutic targets. Although initially focused on Breast Adenocarcinoma, ASCEND is scalable to various cancer types, providing much promise for its future in cancer analysis. Future developments of ASCEND include integration into a user-friendly interface with adjustable parameters and interactive, graphical result displays to maximize transparency and customizability. Overall, ASCEND's dual perspective on cancer analysis provides a comprehensive, precise approach to cancer biomarker identification and epigenetic analysis, facilitating more effective early prevention measures and targeted genomic cancer therapeutics.
Speaker(s):
Jay Ananth, High School
Troy High School

Author(s):
Jay Ananth, High School - Troy High School; Alper Uzun, PhD - Brown, University, Department of Pathology and Laboratory Medicine, Department of Pediatrics;

Exploring Rare Disease Exome-Capture Variants Beyond Canonical Splice Sites with SpliceAI

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

Abstract Keywords: Biomarker Discovery and Development, Genomics/Omic Data Interpretation, Bioimaging Techniques and Applications, Advanced Data Visualization Tools and Techniques, Informatics Research/Biomedical Informatics Research Methods, Proactive Machine Learning and Reinforcement Learning, Machine Learning, Generative AI, and Predictive Modeling, Proactive Machine Learning and Reinforcement Learning
Working Group: Genomics and Translational Bioinformatics Working Group
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Translational Bioinformatics Using Multi-Modal Patient Data and AI

Splice site mutations is critical for producing functional proteins. A number of disease-associated disrupt splicing, disease like spinal muscular atrophy (SMA) is directly linked to splicing mutations. In SMA, a deletion or mutation affects the SMN1 gene's ability to properly splice its RNA, leading to motor neuron loss and hereditary diseases and congenital disease. The application of deep learning tools like SpliceAI, offers an unprecedented opportunity to explore these regions by predicting splice site alterations with high accuracy. While predicting genetic variants using SpliceAI is a time-consuming process, presenting a significant challenge when analyzing the entire spectrum of a patient's variants detected by WES. This study aims to identify and characterize rare disease-associated variants beyond the traditional splice site boundaries using SpliceAI, thereby broadening the scope of genetic diagnostics and therapeutic target identification in rare diseases.
We train a model with positive and negative variant by using HGMD variants of 3,536 donor sites, 1,508 acceptor sites, GenomAD 1,417 donor sites, 2.877 acceptor sites GenomAD datasets, models are basic examination of largest score changes relevant to the nearby splice site, Acceptor model: y = 1 / (1 + e –(3.3235*AG + 6.4756*AL - 2.1955)) with 98% specificity and 0.74 sensitivity. Donor model: y = 1 / ( 1 + e -(3.0560*DG + 7.5341*DL - 1.1142)) • with 88% specificity and 0.89 sensitivity . Neither • y = 1 / ( 1 + e (5.8619*(max(AG, DG)) - 0.7673)). The cutoff is 0.5. Prediction models based on the spliceAI scores can suggest if splicing is altered by weighting splice site gains and losses differently.
Speaker(s):
jinlian wang, PhD
UTHealth

Author(s):

Improving Interoperability of Genomics Data Standards: Evaluation of Alignment Performed Concurrently with Specification Development to Maximize Harmonization and Encourage Convergent Evolution

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

Abstract Keywords: Data Standards, Data Sharing/Interoperability, Genomics/Omic Data Interpretation, Data Transformation/ETL, FHIR
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

Interoperability of genetic data is needed to achieve promises of genomic medicine. However, there is no single standard that conveys all types of genetic data for both clinical and research use cases, and thus interoperability among specifications used by different communities is necessary. This project aims to improve interoperability between two genomic data standards: the HL7 FHIR Molecular Definition resource and the GA4GH Variant Representation Specification. This presentation describes the successes, lessons learned, and opportunities for future work.
Speaker(s):
Aly Khalifa, PhD
Mayo Clinic

Author(s):
Aly Khalifa, PhD - Mayo Clinic; Salem Bajjali, Master of Science - Mayo Clinic; Xianfeng Chen, Ph.D - Mayo Clinic; Sarah Senum, MS - Mayo Clinic; Robert Freimuth, PhD - Mayo Clinic;

Linking mutation burden and somatic mutations signatures to gene expression patterns in the developing human brain

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

Abstract Keywords: Genotype-phenotype Association Studies (including GWAS), Genomics/Omic Data Interpretation, Genomics/Omic Data Interpretation
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Translational Bioinformatics Using Multi-Modal Patient Data and AI

While germline mutations are inherited, somatic mutations can arise during prenatal brain development and cause neurological disease and neurodevelopmental disorders. The rates and patterns of somatic mutations and their relation to gene expression across healthy developing brain regions are poorly studied. In this study, we present a framework to quantify the effects of mutation in general and somatic mutation signatures on gene expression profiles of the human developing brain regions.
Speaker(s):
Judith Somekh, PhD
University of Haifa

Author(s):
Judith Somekh, PhD - University of Haifa; Isana Veksler-Lublinsky, Dr. - Ben-Gurion University of the Negev; Or Amar, Mr. - University of Haifa; Isaac Kohane, MD, PhD - Harvard Medical School;

Integrative Biomedical Informatics in Precision Oncology: Insights from a Head and Neck Cancer Case Study with the NIH All of Us Research Program

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

Abstract Keywords: Informatics Research/Biomedical Informatics Research Methods, Data/System Integration, Standardization and Interoperability, Clinical and Research Data Collection, Curation, Preservation, or Sharing, Cohort Discovery, Data-Driven Research and Discovery, EHR-based Phenotyping, Genotype-phenotype Association Studies (including GWAS)
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

Head and neck cancer (HNC) encompasses a diverse group of malignancies exhibiting a range of etiologies and varying clinical behavior posing clinical challenges, necessitating a comprehensive understanding of the multifactorial nature of HNC. Utilizing computational methods to integrate, transform, and analyze datasets from heterogenous sources, we identify and characterize a HNC cohort revealing correlation across clinical stages using the NIH All of Us Research Program underlying the importance of large-scale data repositories for precision oncology.
Speaker(s):
Sari Mayhue, PhD
Medical University of South Carolina

Author(s):

LUNAR: Genomic Cross-Attention for Early Glioma Recurrence Prediction

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

Abstract Keywords: Machine Learning, Generative AI, and Predictive Modeling, Clinical Genomics/Omics and Interventions Based on Omics Data, Secondary Use of EHR Data
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Translational Bioinformatics Using Multi-Modal Patient Data and AI

Gliomas represent approximately 25.5% of all primary brain and central nervous system (CNS) tumors and 80.8% of malignant brain and CNS tumors. The infiltrative nature and distinct biology of these tumors lend to a high likelihood of eventual cancer recurrence. Insight into a patient’s likelihood of recurrence can profoundly impact patient outcomes; however, there are currently no widely available prediction models for assessing the risk of early glioma recurrence. As such, we developed LUNAR to predict early glioma recurrence using clinical and genomic data from patients with grade II-IV gliomas. Our models achieved AUROCs of 91.15% and 89.1% on The Cancer Genome Atlas and Glioma Longitudinal Analysis datasets."
Speaker(s):
Jessica Patricoski, PhD Candidate, Computational Biology
Brown University Center for Computational Molecular Biology

Author(s):
Jessica Patricoski, PhD Candidate, Computational Biology - Brown University Center for Computational Molecular Biology; Seema Nagpal, MD - Stanford University; Ritambhara Singh, PhD - Brown University; Jeremy Warner, MD, MS - Brown University; Ece Uzun, PhD - Lifespan/Brown University;

Pharmacogenomics Utilization as Prescribing Reassurance: Improving Genomically Compatible Prescribing

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

Abstract Keywords: Pharmacogenomics, Implementation Science and Deployment, Genomics/Omic Data Interpretation
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Integrating Clinical Research and Clinical Care Workflows

This study successfully implemented pharmacogenomics (PGx) clinical decision support (CDS) for inpatients across three institutions. Care teams could optionally utilize PGx CDS during an admission. Our primary end point was to identify differences in prescribing associated with PGx CDS use. We found that care teams who utilized PGx CDS prescribed at higher rates and prescribed more genomically favorable medications compared to those who did not.
Speaker(s):
Zhong Huang, BA
The University of Chicago Pritzker School of Medicine

Author(s):
Zhong Huang, BA - The University of Chicago Pritzker School of Medicine; Matthew Jack - University of Chicago - Center for Personalized Therapeutics; Kevin O'Leary, MD - Northwestern University; Edith Nutescu, PharmD - University of Illinois Chicago; Thomas Chen, PharmD, MD - The University of Chicago; Gregory Ruhnke, MD - The University of Chicago; Randall Knoebel, PharmD - The University of Chicago; Seth Hartman, PharmD - The University of Chicago; Anish Choksi, PharmD - The University of Chicago; Kiang-Teck Yeo, PhD - The University of Chicago; Minoli Perera, PharmD, PhD - Northwestern University; Mark Ratain, MD - The University of Chicago; David Meltzer, MD - The University of Chicago; Peter O'Donnell, MD - The University of Chicago;

Overcoming Challenges in CNV Analysis: Advancing Toward Accurate Identification and Interpretation in Somatic Cancers

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

Abstract Keywords: Genomics/Omic Data Interpretation, Biomarker Discovery and Development, Data Integration, Pharmacogenomics, Patient-centered Research and Care, Bioimaging Techniques and Applications, Data Integration, Informatics Research/Biomedical Informatics Research Methods
Working Group: Genomics and Translational Bioinformatics Working Group
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Implementation Science and Deployment in Informatics: Enabling Clinical and Translational Research

Copy Number Variations (CNVs) in somatic cancers present intricate genomic landscapes that pose significant challenges for accurate identification and interpretation. The accurate calling and curation of CNVs are crucial for understanding tumor biology and informing therapeutic decisions. However, intrinsic complexities within CNVs, as well as technical limitations in estimating key parameters such as tumor purity and ploidy, can lead to erroneous conclusions that impede clinical application.The current landscape of CNV analysis is fraught with difficulties, primarily due to the complexity of CNV profiles in somatic cancers. Inaccurate parameter estimates, such as tumor purity and ploidy, can substantially skew CNV calling algorithms, resulting in unreliable data that can impact clinical decision-making. Additionally, the manual review and reporting of CNVs, especially in regions with complex events such as chromothripsis, can be exceedingly time-consuming and ambiguous.There exists a significant gap in the availability of tools that can efficiently manage the complexity of CNV data while providing accurate and clinically relevant interpretations. The need for a tool that allows for the interactive correction of estimates and manual re-segmentation of calls is evident. Such a tool should not only expedite the review process of CNAs (Copy Number Abnormalities) but also minimize the inherent ambiguities in CNV analysis.
Speaker(s):
Hui Li, Phd
University of Texas Health Science Center at Houston

Author(s):

Analysis of Downregulated and Upregulated Gene Expression and Biological Processes in Healthy and Diseased Kidneys

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

Abstract Keywords: Single Cell Analysis, Ontologies, Advanced Data Visualization Tools and Techniques
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Real-World Evidence in Informatics: Bridging the Gap between Research and Practice

The Kidney Precision Medicine Project (KPMP) is a multi-year effort to understand kidney disease, focusing on acute kidney injury (AKI) and chronic kidney disease (CKD). AKI occurs when the kidneys suddenly stop working properly, from a minor loss of kidney function to complete kidney failure, typically as a complication of another illness. CKD develops slowly over time due to kidney damage, and results in impaired kidney function, while showing few symptoms initially. CKD increases the risk of other health problems like heart disease and stroke. In this study, we are reanalyzing gene expression data generated by KPMP using Gene Ontology-based term enrichment to understand how biological processes are altered in AKI and CKD versus normal controls. We focused on kidney-specific genes and our results primarily showed changes in the regulation of kidney-specific processes, typically in transport, metabolism, or kidney development GO terms. For instance, we saw downregulation of transmembrane transport, organic acid transport, tube development, and kidney development in cell types like epithelial cells of proximal tubule. On the other hand, in CKD, we saw upregulation in renal tubule development, kidney morphogenesis, and thick ascending limb development in the kidney loop of Henle thin descending epithelial cell. Our results show that kidney processes are differentially affected in AKI and CKD. Due to our focus on kidney-specific genes, our data did not identify immune system processes that are downregulated or upregulated. Our future work will look at genes associated with immune system processes that are differentially expressed in diseased kidney.
Speaker(s):
Author(s):
Jiaxing Liu, MS - The University at Buffalo; Edy Munoz, BS - University at Buffalo; Skyler Resendez - The University at Buffalo; Alexander Diehl, PhD - University at Buffalo; Yongqun He, PhD - University of Michigan;

NET CANCER INSIGHT: Cancer Network Annotation and Comparative Analysis Tool

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

Abstract Keywords: Systems Biology and Network Analysis, Data Integration, Biomarker Discovery and Development, Data-Driven Research and Discovery, Ontologies
Primary Track: Translation Bioinformatics/Precision Medicine
Programmatic Theme: Novel Methods for Variant Detection and Interpretation from Omics Data

Cancer is a multifaceted disease driven by complex protein-protein interactions (PPI). Mapping these interactions as networks helps identify hub proteins and driver genes, offering insights into cancer’s genetic underpinnings. To aid this effort, we present Net Cancer Insight, a computational tool designed for the visualization, annotation, and comparison of cancer PPI networks. Developed using React JS and leveraging npm modules such as d3 and recharts, Net Cancer Insight integrates data from trusted databases including OncoKB, NCG, Reactome, GO terms, STRING, and cBioPortal. The tool generates interactive network graphs, allowing for detailed network analyses, including graph theory metrics, shared protein identification, and separation score analysis. Users can filter queries, load backend data, and identify interactions across major databases. Net Cancer Insight offers a comprehensive solution for exploring cancer's genetic landscape and is freely available at https://github.com/alperuzun/NetCancer-Insight.
Speaker(s):
Alper Uzun, PhD
Brown, University, Department of Pathology and Laboratory Medicine, Department of Pediatrics

Author(s):