A Retrospective Cohort Study: Predicting 90-Day Mortality for ICU Trauma Patients with a Machine Learning Algorithm
Shan Yang1, Lirui Cao2, Yongfang Zhou3
1Department of Critical Care Medicine, West China Hospital of Sichuan University, Chengdu, Sichuan, 610041, People's Republic of China.
Journal of Multidisciplinary Healthcare
|September 13, 2023
Summary
This study developed an XGBoost model to predict 90-day mortality in intensive care unit (ICU) trauma patients. The model demonstrated superior performance, aiding early risk identification and intervention.
Area of Science:
- Medical Informatics
- Machine Learning in Healthcare
- Trauma Care Research
Background:
- Predicting mortality in ICU trauma patients is crucial for timely intervention.
- Machine learning offers potential for improving mortality prediction accuracy.
Purpose of the Study:
- To develop and validate a machine learning model for predicting 90-day mortality in ICU trauma patients.
- To compare the performance of various machine learning algorithms for this prediction task.
Main Methods:
- Utilized data from the MIMIC-III database for severe trauma patients.
- Developed and evaluated nine machine learning models, including XGBoost, logistic regression, and random forest.
- Assessed model performance using discrimination, calibration, and clinical application metrics.
Main Results:
- The XGBoost model achieved high accuracy (82.8%), sensitivity (79.7%), and specificity (77.6%).
- XGBoost outperformed the other eight machine learning models in predicting 90-day mortality.
- The XGBoost model demonstrated favorable calibration compared to logistic regression.
Conclusions:
- The XGBoost model is a robust tool for predicting 90-day mortality in ICU trauma patients.
- This model can assist clinicians in early risk factor identification and intervention strategies.
- Machine learning, particularly XGBoost, shows significant promise in enhancing trauma patient outcomes.
