Machine Learning Prediction Models for Postoperative Stroke in Elderly Patients: Analyses of the MIMIC Database

Area of Science:

• Geriatric Medicine
• Neurology
• Data Science

Background:

• Postoperative stroke is a growing concern in aging populations.
• Accurate prediction and risk factor identification are crucial for elderly patients.

Purpose of the Study:

• To develop a reliable prediction model for postoperative stroke in elderly patients.
• To identify key risk factors contributing to postoperative stroke.

Main Methods:

• Utilized machine learning (ML) models on MIMIC-III and MIMIC-VI databases.
• Applied SMOTENC for data balancing and iterative SVD for imputation.
• Evaluated seven modeling approaches, including XGBoost, using ROC curves.

Main Results:

• The XGBoost model achieved the highest Area Under the Curve (AUC) of 0.78.
• Identified hypertension, cancer, congestive heart failure, chronic pulmonary disease, and peripheral vascular disease as top predictors.
• Hypertension demonstrated significant predictive value for postoperative stroke.

Conclusions:

• Postoperative stroke in elderly patients is reliably predictable.
• The XGBoost model is a robust predictive tool.
• History of hypertension is a critical factor for preventing postoperative stroke, outweighing laboratory tests.