Publication Date

2025

Document Type

Thesis

Committee Members

Tanvi Banerjee, Ph.D. (Advisor); Timothy T. Phamduy, D.O. (Committee Member); Tomojit Ghosh, Ph.D. (Committee Member)

Degree Name

Master of Science (MS)

Abstract

Sepsis is a leading cause of pediatric mortality, claiming more lives in the United States annually than all childhood cancers combined. Early identification in Emergency Departments (EDs) remains challenging, as the current Phoenix criteria establishes an updated international consensus definition for sepsis, however is not designed for use as a screening tool. This study aimed to develop predictive models identifying pediatric patients at risk of sepsis within 24 hours of admission. Multiple tree-based and deep learning models were trained utilizing clinical and laboratory data from the initial four hours of presentation. Both the LightGBM and LSTM architectures demonstrated superior performance, with the LightGBM model achieving an AUC of 0.93 and a recall of 0.96. Predictions were primarily driven by dynamic changes in mean arterial pressure, respiratory rate, and heart rate. Furthermore, unsupervised K-means clustering identified two clinically distinct subgroups within the sepsis cohort, exhibiting phenotypes consistent with early Sepsis-Induced Coagulopathy (SIC) and critical Overt Disseminated Intravascular Coagulation (DIC). These findings highlight the potential of advanced machine learning for both early prediction and physiological stratification in pediatric sepsis management.

Page Count

84

Department or Program

Department of Computer Science and Engineering

Year Degree Awarded

2025


Share

COinS