<![CDATA[Heart disease remains one of the leading causes of mortality worldwide, highlighting the need for early prediction methods capable of accurately identifying individuals at risk. This study aims to develop a heart disease prediction model based on Random Forest by integrating a Genetic Algorithm for feature selection and SMOTEENN to address class imbalance. The research stages include data preprocessing, class balancing using SMOTEENN, feature selection using a Genetic Algorithm, Random Forest model development, and hyperparameter optimization using GridSearchCV, with recall as the primary evaluation metric. The initial Random Forest model demonstrated good predictive performance, achieving a recall of 0.94 for the positive class on the test set, while yielding a lower recall of 0.74 for the negative class, indicating an imbalance in classification performance across classes. Following the application of SMOTEENN, Genetic Algorithm-based feature selection, and hyperparameter optimization, the optimized model achieved a mean recall of 0.91 based on cross-validation. These findings indicate that the proposed approach maintains high predictive sensitivity while incorporating mechanisms to address class imbalance and identify relevant features. Therefore, the integration of Random Forest, SMOTEENN, and Genetic Algorithm demonstrates potential as a predictive approach to support the early detection of heart disease. However, as this study employs a non-local dataset, further validation using datasets representative of the target population is required before the model can be generalized or applied to specific populations.]]>
Copyrights © 2026
