Machine learning-based mortality prediction models for emergency department patients - a comparative analysis

Aug 5, 2026

Upload by Aurelia

Accurate in-hospital mortality prediction in emergency departments (ED) is essential for timely intervention and efficient resource allocation. This study retrospectively analyzed 1,389 ED patients admitted to Affiliated Kunshan Hospital of Jiangsu University in 2021, deriving demographic data, vital signs, and laboratory results within 30 minutes of arrival. Nine machine learning models were trained and compared using AUROC, sensitivity, specificity, and calibration. LightGBM achieved the highest discrimination (AUROC 0.9605), with SHAP highlighting serum lactate, GCS, albumin, base excess, and systolic blood pressure as key predictors, and decision curve analysis supporting clinical utility.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Introduction

Emergency department mortality prediction and triage

Machine learning methods in predictive modeling

Existing gaps and study motivation

OPEN ACCESS EDITED BY Miodrag Zivkovic, Singidunum University, Serbia REVIEWED BY Nebojsa Bacanin, Singidunum University, Serbia Oral Menteş, Gülhane Askerî Tıp Akademisi, Türkiye *CORRESPONDENCE Hua Yuan [yeking3@hotmail.com](yeking3@hotmail.com) †These authors have contributed equally to this work RECEIVED 09 October 2025 REVISED 14 January 2026 ACCEPTED 16 January 2026 PUBLISHED 30 January 2026 CITATION Jiang Z, Ma J, Guo Z, Feng Q and Yuan H (2026) Machine learning-based mortality prediction models for emergency department patients: a comparative analysis. Front. Med. 13:1721101 . doi: 10.3389/fmed.2026.1721101 COPYRIGHT © 2026 Jiang, Ma, Guo, Feng and Yuan. 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. TYPE Original Research PUBLISHED 30 January 2026 DOI 10.3389/fmed.2026.1721101 Machine learning-based mortality prediction models for emergency department patients: a comparative analysis Zhen Jiang†, Jin Ma†, Zhiqiang Guo, Qiupeng Feng and Hua Yuan * Department of Emergency Medicine, Affiliated Kunshan Hospital of Jiangsu University, Kunshan, China Background: Accurate mortality prediction in emergency departments (ED) is crucial for timely intervention and resource allocation. This study developed and compared multiple machine learning models to predict in-hospital mortality among ED patients. Methods: We retrospectively analyzed 1,389 ED patients admitted to Affiliated Kunshan Hospital of Jiangsu University between January and December 2021. After excluding patients under 16 years and those transferred or discharged against medical advice, we collected demographic data, vital signs, and laboratory results within 30 min of ED arrival. Nine machine learning models including Logistic Regression, Random Forest, XGBoost, LightGBM, Gradient Boosting, Support Vector Machine (SVM), Neural Network, AdaBoost, and an ensemble voting classifier were developed and compared using metrics including area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, and calibration. Results: Among 1,389 patients (mean age 67.72 ± 19.28 years, 63. 1% male), the mortality rate was 11. 59% . LightGBM demonstrated the best performance with an AUROC of 0.9605 (95% CI: 0.94–0.98), sensitivity of 78. 12%, and specificity of 93.90% . The ensemble voting classifier achieved comparable performance (AUROC: 0.9599) . SHAP analysis identified serum lactate (importance: 0. 252), Glasgow Coma Scale (GCS) (0 .085), albumin (0 .075), base excess (BE) (0 .061), and systolic blood pressure (SBP) (0 .049) as the top five predictive features. Calibration curves demonstrated excellent agreement between predicted and observed mortality rates, and decision curve analysis confirmed clinical utility across various threshold probabilities. Risk stratification based on predicted mortality probabilities effectively separated patients into prognostically distinct groups. Conclusion: Machine learning models, particularly LightGBM, provide highly accurate mortality prediction for ED patients. The integration of readily available clinical and laboratory parameters enables early risk stratification and may facilitate targeted interventions to improve patient outcomes. KEYWORDS emergency department, LightGBM, machine learning, mortality prediction, risk stratification, SHAP analysis Frontiers in Medicine 01 [frontiersin.org](frontiersin.org) 1 Introduction Emergency departments serve as critical entry points for acutely ill patients, where rapid assessment and appropriate triage decisi

FAQs

Why is mortality prediction important in emergency departments?

Accurate early mortality risk estimation supports timely intervention, better resource allocation, and informed clinical decision-making during triage.

What data and timeframe were used to build the predictive models?

The study used demographic data, vital signs, and laboratory results collected within 30 minutes of ED arrival, from 1,389 eligible adult patients.

Which model performed best and what were the top predictive features?

LightGBM showed the best performance with an AUROC of 0.9605. SHAP identified serum lactate, Glasgow Coma Scale (GCS), albumin, base excess (BE), and systolic blood pressure (SBP) as the top features.

Aug 5, 2026

Upload by Aurelia