Description

An onsite recording of this session will be included in the Symposium OnDemand offering. 

Sepsis Prediction Models are Trained on Labels that Diverge from Clinician-Recommended Treatment Times

Presentation Time: 03:30 PM - 03:45 PM

Abstract Keywords: Clinical Decision Support, Critical Care, Internal Medicine or Medical Subspecialty, Infectious Diseases and Epidemiology
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Many sepsis prediction models use the Sepsis-3 definition or its variants as a training label. However, among the few sepsis models ever deployed in practice, there is scant evidence that they offer clinically meaningful decision support at the bedside. As a potential mechanism to explain this limitation, we hypothesized that clinician-recommended treatment times for sepsis would diverge from onset time defined by Sepsis-3. We conducted an electronic survey that was completed by 153 clinicians at three large and geographically diverse medical centers using vignettes derived from eight real cases of sepsis. After reviewing these vignettes, participants suggested antibiotic treatment to start an average of 7.0 hours (95% confidence interval 5.3 to 8.8) before the Sepsis-3 definition onset. Thus, predicting Sepsis-3 onset as a treatment prompt could lead to inappropriate and delayed treatment recommendations. Building predictive decision support systems that identify outcomes aligned with bedside decisions would increase their clinical utility.

Speaker(s):
GARY WEISSMAN, MD, MSHP
University of Pennsylvania

Author(s):
GARY WEISSMAN, MD, MSHP - University of Pennsylvania; Rebecca Hubbard, PhD - University of Pennsylvania; Blanca Himes, PhD - University of Pennsylvania; Kelly Goodman-O'Leary, MSN, CRNP - PAIR Center, University of Pennsylvania; Michael Harhay, PhD - University of Pennsylvania Perelman School of Medicine; Jennifer Ginestra, MD, MSHP - University of Pennsylvania Perelman School of Medicine; Rachel Kohn, MD, MSCE - University of Pennsylvania Perelman School of Medicine; Andrew Admon, MD, MPH - University of Michigan; Stephanie Parks Taylor, MD - Atrium Health; Scott Halpern, MD, PhD - University of Pennsylvania;

Constructing a Cerebral Hemodynamics Model within a Data Assimilation Pipeline to Enhance Clinical Decision Support in Neurocritical Care

Presentation Time: 03:45 PM - 04:00 PM

Abstract Keywords: Critical Care, Simulation of Complex Systems, Clinical Decision Support, Biomarkers, Precision Medicine, Machine Learning, Computational Biology, Bioinformatics
Primary Track: Foundations
Programmatic Theme: Clinical Research Informatics

Brain injuries impact cerebral blood flow and are a major cause of death. Forecasts of cerebral blood flow and measurement of the functionality of mechanisms that regulate blood flow (CVTR) are not possible but could be transformative for neurocritical care. We use a new model of cerebral hemodynamics estimated with data assimilation to compute CVTR functionality and forecast CBF, offering promising prospects for transforming clinical decision support and improving patient outcomes.

Speaker(s):
Jennifer Briggs, B.S.
University of Colorado Anschutz Medical Campus

Author(s):
David Albers, PhD - University of Colorado, Department of Biomedical Informatics; J.N. Stroh, PhD - University of Colorado Anschutz Medical Campus; Soojin Park, MD - Columbia University Medical Center; Tellen Bennett, MD, MS - University of Colorado School of Medicine; Brandon Foreman, MD MS FACNS FNCS - University of Cincinnati Medical Center/University of Cincinnati Gardner Neuroscience Institute;

Neural Granger Causal Discovery for Derangements in ICU-Acquired Acute Kidney Injury Patients

Presentation Time: 04:00 PM - 04:15 PM

Abstract Keywords: Clinical Decision Support, Data Mining, Causal Inference, Information Extraction, Machine Learning
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Nowadays, healthcare systems increasingly utilize automated surveillance of electronic medical record (EMR) data to detect adverse events with specific patterns. Despite these technological advances, the early identification of adverse events remains challenging due to the absence of clearly defined prodromal sequences that could signal the onset of such events. Achieving clinically meaningful and interpretable prediction outcomes necessitates a framework that is capable of (i) deducing temporal relationships among various time series features within EMR data (e.g., laboratory test results, vital signs), and (ii) identifying specific patterns that herald the occurrence of an adverse event (e.g., acute kidney injury (AKI)). This study employs a time series forecasting approach to undertake neural Granger causal analysis, and further enhance it by integrating a personalized PageRank algorithm to analyze the critical causal derangements among ICU-acquired AKIpatients. An experimental analysis based on the proposed methodology was conducted using a dataset from MIMIC-IV.Finally, a Granger causality (GC) graph, which revealed several interpretable GC chains that could be used to predict the occurrence of AKI in ICU settings, was generated. The GC graph and GC chains identified in this study have the potential to aid ICU physicians in providing timely interventions and may help improve patient outcomes.

Speaker(s):
Guilan Kong, PhD
National Institute of Health Data Science, Peking University

Author(s):
Haowei Xu, Ms. - School of Artificial Intelligence, Optics and Electronics (iOPEN), Northwestern Polytechnical University; Wentie Liu, Ms. - National Institute of Health Data Science, Peking University; Tongyue Shi, Ms. - National Institute of Health Data Science, Peking University;

Identifying acute kidney injury subtypes based on serum electrolyte data in ICU via K-medoids clustering

Presentation Time: 04:15 PM - 04:30 PM

Abstract Keywords: Precision Medicine, Data Mining, Machine Learning
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This study proposes to use the K-medoids clustering method to identify subtypes of Intensive Care Unit (ICU)-acquired acute kidney injury (AKI) patients based on serum electrolyte data. Three distinct AKI subtypes with different serum electrolyte characteristics were identified by clustering analysis. Further, descriptive analysis was employed to characterize in-hospital mortality and renal replacement therapy, diuretic and vasopressor usage in the three subtypes, and Chi-square tests were conducted to check the differences of prognosis and treatments among the identified subtypes. This study enables the subclassification of AKI patients in the ICU, facilitating ICU physicians to make timely clinical decisions about AKI, and ultimately may contribute to patient outcome improvement.

Speaker(s):
Guilan Kong, PhD
National Institute of Health Data Science, Peking University

Author(s):

Predicting Prolonged Respiratory Failure via Unsupervised and Supervised Machine Learning Models

Presentation Time: 04:30 PM - 04:45 PM

Abstract Keywords: Critical Care, Machine Learning, Clinical Decision Support
Primary Track: Applications

Predicting prolonged respiratory failure (PRF, patients requiring intubation ≥14 days for this task) in patients requiring mechanical ventilation is challenging but beneficial for treatment planning. This study applies supervised and unsupervised machine learning methods on an ICU cohort with confirmed/suspected pneumonia for PRF prediction. Group-based multivariate trajectory modeling (GBMT) on five ventilator parameters collected during the first five days of intubation identified four groups that represent unique phenotypes and are predictive of PRF. Multivariate Time Series Transformer (MVTS) showed superior discrimination of PRF using time-series clinical data in comparison with XGBoost on baseline data.

Speaker(s):
Yanyi Ding, MS
Northwestern University

Author(s):
Yanyi Ding, MS - Northwestern University; Meghan Hutch, BS - Northwestern University - Feinberg School of Medicine; Thomas Stoeger, PhD - Northwestern University; Yuan Luo, PhD - Northwestern University; Catherine Gao, MD - Northwestern;

Influence of the CONCERN Early Warning System on Unanticipated ICU Transfers, In-Hospital Mortality, and Length of Stay: Results from a Multi-site Pragmatic Randomized Controlled Clinical Trial

Presentation Time: 04:45 PM - 05:00 PM

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

Communicating Narrative Concerns Entered by RNs Early Warning System (CONCERN EWS) is a machine-learning predictive model that leverages nursing surveillance documentation patterns to predict deterioration risks for hospitalized patients. In a retrospective cohort study of 1,013 hospital encounters with unanticipated ICU transfers from a multi-site pragmatic randomized controlled trial, we assessed the influence of CONCERN EWS on in-hospital mortality and length of stay following unanticipated ICU transfers. Chi-square tests, t-tests, multivariate logistic regression, and generalized linear models were used. Our findings showed that patients who had unanticipated ICU transfers from acute care units with CONCERN EWS had a lower in-hospital mortality rate and a shorter average hospital stay than those transferred from units receiving usual care. These results suggest that CONCERN EWS significantly enhances situational awareness for care teams, improves communication, and effectively facilitates timely interventions, thereby streamlining care processes and improving patient outcomes.

Speaker(s):
Rachel Lee, PhD, RN
Columbia University

Author(s):
Kenrick Cato, PhD, RN, CPHIMS, FAAN - University of Pennsylvania/ Children's Hospital of Philadelphia; Patricia Dykes, PhD, RN, FAAN, FACMI - Brigham and Women's Hospital, Harvard Medical School; Graham Lowenthal, BA - Brigham and Women's Hospital; Jennifer Withall, PhD - Columbia University Department of Biomedical Informatics; Sandy Cho, Nurse Director - Newton-Wellesley Hospital; Haomiao Jia, PhD - Columbia University School of Nursing; Sarah Rossetti, RN, PhD - Columbia University Department of Biomedical Informatics;