Interpretable Machine Learning-Based Prediction of 28-Day Mortality in ICU Patients With Sepsis - A Multicenter Retrospective Study

Aug 5, 2026

Sepsis remains a major cause of ICU mortality and a persistent global health challenge. This multicenter retrospective study develops and evaluates a machine learning model to forecast 28-day all-cause mortality in ICU patients with sepsis. Data were drawn from the eICU Collaborative Research Database and predictors were selected using LASSO regression and Boruta feature selection. Model performance was assessed via multiple classification metrics with 10-fold cross-validation, and interpretability analysis was performed on the most stable model to support clinical insight.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Introduction

Background and clinical problem

Methods

Data source and outcomes

Feature selection and model building

Performance evaluation and interpretability

Results

Cohort description

Model comparison and key predictors

Conclusion

TYPE Original Research PUBLISHED 08 January 2025 DOI 10.3389/fcimb.2024.1500326 OPEN ACCESS EDITED BY Gang Ye, Sichuan Agricultural University, China REVIEWED BY Nozomi Takahashi, University of British Columbia, Canada Kawther Alquadan, University of Florida, United States Bin Yi, Army Medical University, China Jie Weng, Second Afﬁliated Hospital and Yuying Children ’s Hospital of Wenzhou Medical University, China *CORRESPONDENCE Zhiping Li [zpli@fudan.edu.cn](zpli@fudan.edu.cn)[ ](zpli@fudan.edu.cn)Yi Wang [yiwang@shmu.edu.cn](yiwang@shmu.edu.cn) †These authors share ﬁrst authorship RECEIVED 23 September 2024 ACCEPTED 16 December 2024 PUBLISHED 08 January 2025 CITATION Shen L, Wu J, Lan J, Chen C, Wang Y and Li Z (2025) Interpretable machine learning-based prediction of 28-day mortality in ICU patients with sepsis: a multicenter retrospective study. Front. Cell. Infect. Microbiol. 14:1500326 . doi: 10.3389/fcimb.2024.1500326 COPYRIGHT © 2025 Shen, Wu, Lan, Chen, Wang and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Interpretable machine learningbased prediction of 28-day mortality in ICU patients with sepsis: a multicenter retrospective study Li Shen 1,2†, Jiaqiang Wu 3†, Jianger Lan 1†, Chao Chen 4, Yi Wang 5* and Zhiping Li 1* 1 Department of Clinical Pharmacy, Children’s Hospital of Fudan University, National Children’s Medical Center, Shanghai, China, 2 Department of Pharmacy, Suzhou Hospital, Afﬁliated Hospital of Medical School, Nanjing University, Suzhou, Jiangsu, China, 3School of Life Sciences and Biopharmaceutical Science, Shenyang Pharmaceutical University, Shenyang, China, 4 Department of Neonatology, Children ’s Hospital of Fudan University, National Children ’s Medical Center, Shanghai, China, 5 Department of Neurology, Children ’s Hospital of Fudan University, National Children ’s Medical Center, Shanghai, China Background: Sepsis is a major cause of mortality in intensive care units (ICUs) and continues to pose a signiﬁcant global health challenge, with sepsis-related deaths contributing substantially to the overall burden on healthcare systems worldwide. The primary objective was to construct and evaluate a machine learning (ML) model for forecasting 28-day all-cause mortality among ICU sepsis patients. Methods: Data for the study was sourced from the eICU Collaborative Research Database (eICU-CRD) (version 2 . 0) . The main outcome was 28-day all-cause mortality. Predictor selection for the ﬁnal model was conducted using the least absolute shrinkage and selection operator (LASSO) regression analysis and the Boruta feature selection algorithm. Five machine learning algorithms including logistic regression (LR), decision tree (DT), extreme gradient boosting (XGBoost), support vector machine (SVM), and light gradient boosting machine (lightGBM) were employed to construct models using 10-fold cross-validation. Model performance was evaluated using AUC, accuracy, sensitivity, speciﬁcity, recall, and F1 score. Additionally, we performed an interpretability analysis on the model that showed the most stable performance. Results: The ﬁnal study cohort comprised 4564 patients, among whom 568 (12 . 4%) died within 28 days of ICU admission . The XGBoost algorithm demonstrated the most reliable performance, achieving an AUC of 0.821, balancing sensitivity (0 . 703) and speciﬁcity (0 . 798) . The top three risk predictors of mortality included APACHE score, serum lactate levels, and AST. Frontiers in Cellular and Infection Microbiology 01 [frontiersin.org](

FAQs

What is the main objective of the study?

To construct and evaluate a machine learning model that forecasts 28-day all-cause mortality among ICU sepsis patients.

Which dataset and outcome were used?

The study used data from the eICU Collaborative Research Database (version 2.0). The primary outcome was 28-day all-cause mortality.

How were features selected and models validated?

Predictors were selected using LASSO regression and the Boruta feature selection algorithm. Five ML algorithms were trained with 10-fold cross-validation and evaluated using AUC, accuracy, sensitivity, specificity, recall, and F1 score.

Aug 5, 2026

Upload by Theodore