Predicting In-Hospital Mortality in Patients With Acute Myocardial Infarction: A Comparison of Machine Learning
Hamidreza Soleimani1, Soroush Najdaghi2, Delaram Narimani Davani2
1Tehran Heart Center, Cardiovascular Disease Research Institute, Tehran University of Medical Sciences, Tehran, Iran.
Clinical Cardiology
|March 27, 2025
Summary
Machine learning accurately predicts in-hospital mortality in acute myocardial infarction (AMI) patients. Key predictors include reduced left ventricular ejection fraction (LVEF) and elevated fasting blood glucose, aiding clinical decisions.
Area of Science:
- Cardiology
- Medical Informatics
- Machine Learning
Background:
- Acute myocardial infarction (AMI) is a major global cause of death.
- Predicting in-hospital mortality in AMI patients is crucial for timely intervention.
- Advanced machine learning (ML) offers novel approaches to identify mortality predictors.
Purpose of the Study:
- To explore predictors of in-hospital mortality among AMI patients.
- To evaluate the performance of various ML algorithms in predicting AMI mortality.
- To identify key clinical, demographic, and laboratory variables associated with AMI mortality.
Main Methods:
- Analysis of 7422 AMI patients treated with percutaneous coronary intervention (PCI).
- Evaluation of 58 clinical, demographic, and laboratory variables using seven ML algorithms (RF, LASSO, XGBoost).
- Utilized SMOTE for class imbalance, cross-validation, and SHAP for model interpretation.
Main Results:
- Random Forest (RF) demonstrated the highest predictive performance (AUC 0.924).
- Key predictors identified: lower LVEF, higher fasting blood glucose, elevated creatinine, and advanced age.
- Lower LDL-C was also a predictor, while BMI showed no significant association.
Conclusions:
- ML algorithms, especially RF, are effective in predicting in-hospital mortality in AMI patients.
- Identified predictors like LVEF and biochemical markers can enhance clinical decision-making.
- These findings can contribute to improved patient outcomes in AMI care.
