Development and Validation of a Quick Sepsis-Related Organ Failure Assessment-Based Machine-Learning Model for Mortality Prediction in Patients with Suspected Infection in the Emergency Department
Problem
The qSOFA score has limited accuracy for predicting mortality in emergency-department patients with suspected infection.
Why it matters
Accurate early mortality risk prediction in the ED guides timely treatment of suspected infection.
What the AI did
Machine-learning models (XGBoost, LightGBM, random forest) were trained on qSOFA variables from 19,353 patients across three Korean hospitals and tested on 4,234 patients to predict three-day mortality.
Results
AUROC comparison via DeLong's test performed; values not stated in provided abstract
Impact
Large multi-hospital retrospective study of ML mortality prediction; comparative performance not shown in abstract.
Record
| Venue | Journal of Clinical Medicine |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Young Suk Kwon, Moon Seong Baek |
| Citations | 29 (OpenAlex, at import time) |
| References | 47 |
| Model or method | Extreme gradient boosting, LightGBM, random forest |
| Organization | not stated in source |
| Primary source | https://doi.org/10.3390/jcm9030875 |
| Open access | gold |
| Keywords | qSOFA, machine learning, mortality prediction, emergency department, sepsis, XGBoost |
Importance rating
real but narrower advance
Improving mortality risk prediction for suspected infection in the ED is a real but narrower advance over qSOFA; abstract gives no population-scale impact figures.