Related Experiment Videos
Development and Validation of a Machine Learning-Based Risk Assessment Tool for In-Hospital Mortality in Elderly
Yuchen Zhou1, Xinhe Zhou2, Xiaozhu Liu2
1Department of Surgical Intensive Care Unit, Beijing Shijitan Hospital, Capital Medical University, No. 10, Tieyi Road, Haidian District, Beijing 100038, China.
Journal of Clinical Medicine
|June 26, 2026
Summary
A new machine learning model accurately predicts in-hospital mortality in elderly patients experiencing hypoxemia after non-cardiac surgery. This tool aids early risk assessment for better patient outcomes.
Area of Science:
- Medical Informatics
- Machine Learning in Healthcare
- Critical Care Medicine
Background:
- Postoperative hypoxemia is a common complication after non-cardiac surgery.
- It is linked to increased mortality in elderly patients.
- A specific predictive tool for this group is lacking.
Purpose of the Study:
- To develop and validate a machine learning model for predicting in-hospital mortality.
- The focus is on elderly patients with hypoxemia post-non-cardiac surgery.
Main Methods:
- Retrospective cohort study using the MIMIC-IV database.
- Included patients aged 65+ with hypoxemia (PaO2/FiO2 < 300 mmHg) within 48h of ICU admission.
- XGBoost model developed and validated; SHAP used for interpretability.
Main Results:
- 1838 of 6051 (30.4%) patients died during hospitalization.
- XGBoost model achieved AUC 0.794, specificity 0.917, accuracy 0.769.
- Key predictors: vasopressor use, advanced age, and PaO2/FiO2 ratio.
Conclusions:
- The XGBoost model accurately predicts in-hospital mortality in this high-risk group.
- This provides a valuable tool for early risk stratification.
- Facilitates timely interventions for elderly patients with postoperative hypoxemia.
