Prediction of the Fatal Acute Complications of Myocardial Infarction via Machine Learning Algorithms

Area of Science:

• Artificial Intelligence
• Data Science
• Cardiology

Background:

• Myocardial infarction (MI) is a leading cause of mortality, especially within the first year post-event.
• Early identification of patients at high risk of fatal complications is crucial for timely intervention.
• Machine learning (ML) offers powerful tools for analyzing complex health data to predict outcomes.

Purpose of the Study:

• To develop and evaluate a machine learning model for predicting fatal complications of MI within the initial 72 hours of hospital admission.
• To identify key clinical features that contribute to the prediction of adverse outcomes in MI patients.

Main Methods:

• Utilized a database of MI patient demographics and clinical records, classifying outcomes as deceased or alive.
• Applied Recursive Feature Elimination (RFE) for feature selection, reducing the dataset to 50 key features.
• Evaluated four ML algorithms (logistic regression, SVM, random forest, XGBoost) using eight performance metrics.

Main Results:

• The study included 1699 MI patients, with 15.94% experiencing fatal complications.
• The Extreme Gradient Boosting (XGBoost) model demonstrated superior performance, achieving high accuracy (91.47%) and F1-score (95.14%).
• Cardiogenic shock was identified as the most significant predictor of fatal complications in the XGBoost model.

Conclusions:

• XGBoost algorithms show significant promise as a predictive tool for fatal complications following myocardial infarction.
• This ML approach can aid clinicians in risk stratification and timely management of high-risk MI patients.