Using Machine Learning Models to Predict In-Hospital Mortality for ST-Elevation Myocardial Infarction Patients

Area of Science:

• Cardiology
• Medical Informatics
• Machine Learning

Background:

• Acute myocardial infarction, particularly ST-elevation myocardial infarction (STEMI), is a significant cause of death and hospitalization in China.
• Effective prediction of in-hospital mortality for STEMI patients is crucial for timely clinical interventions.

Purpose of the Study:

• To develop accurate and interpretable machine learning models for predicting in-hospital mortality in STEMI patients.
• To leverage the Chinese Acute Myocardial Infarction (CAMI) registry data for model development.

Main Methods:

• Cohort construction and feature engineering were performed on the CAMI registry data.
• Supervised learning methods with high interpretability, including generalized linear models, decision trees, and Bayes models, were employed.
• Models were trained and validated to predict in-hospital mortality.

Main Results:

• The developed interpretable machine learning models demonstrated high prediction performance, with an Area Under the Curve (AUC) ranging from 0.80 to 0.85.
• These models outperformed previous prediction models for STEMI in-hospital mortality.
• The models provide easily interpretable insights for clinical decision support.

Conclusions:

• Interpretable machine learning offers a powerful approach for predicting in-hospital mortality in STEMI patients.
• The developed models can aid clinicians in making informed decisions for STEMI patient management.
• This study highlights the utility of registry data and advanced analytical techniques in improving cardiovascular care.