Improving cardiovascular disease prediction: An optimized tree-based strategy using SMOTE

The heart is a wellspring of health and any related issues can result in serious cardiovascular conditions that lead ultimately to fatality. In this paper, we utilized three robust tree-based machine learning models, the Random Forest, Gradient Boosting Classifier, and XGBoost, to predict the presence or absence of heart disease. The dataset used is a combination of five heart datasets: Cleveland (303 instances), Hungarian (294 instances), Switzerland (123 instances), Long Beach VA (200 instances), and Stalog (Heart) Data Set (270 instances). This dataset presented a moderate imbalance in class distribution. To tackle this issue, we performed the Synthetic Minority Over-sampling Technique (SMOTE). The results indicate that the Gradient Boosting Classifier achieved the highest accuracy among the implemented models.