PERFORMANCE COMPARISON OF LOGISTIC REGRESSION, RANDOM FOREST, AND XGBOOST WITH SMOTE AND HYPERPARAMETER TUNING IN THE CLASSIFICATION OF ACUTE MYOCARDIAL INFARCTION

Abstract

Acute myocardial infarction (AMI) is one of the leading causes of death from cardiovascular disease worldwide. Delayed treatment can lead to various complications ranging from heart failure to sudden death; therefore, early detection is essential to reduce mortality from AMI. This study aims to conduct a comparative analysis of the Logistic Regression, Random Forest, and XGBoost models for AMI classification. To address the issue of class imbalance in medical data and optimize model performance, the Synthetic Minority Oversampling Technique (SMOTE) was applied, along with hyperparameter tuning using Randomized Search Cross-Validation. Model performance was evaluated using the Area Under the Curve (AUC), recall, precision, F1-score, and accuracy. The results of the study show that the application of SMOTE and hyperparameter tuning has varying effects, particularly on Logistic Regression, which experienced a significant improvement. XGBoost demonstrated stable performance across all scenarios, while Random Forest with SMOTE produced the highest performance and the lowest number of prediction errors, achieving a recall of 1.000, precision of 0.992, and ROC-AUC of 0.999. Thus, XGBoost excels in terms of stability, while Random Forest is more accurate and reliable in terms of minimizing prediction errors in the classification of Acute Myocardial Infarction.

Keywords: Acute Myocardial Infarction (AMI), Extreme Gradient Boosting, Hyperparameter Tuning, Logistic Regression, Random Forest, Synthetic Minority Oversampling Technique (SMOTE)