<![CDATA[Heart disease remains a major health problem worldwide. The World Health Organization (WHO) reports that in 2022, approximately 19.8 million people died from heart disease, highlighting the need for the implementation of an appropriate early detection model. This study proposes a hybrid SVM–PSO model with correlation-based feature selection, duplicate data handling, and a multi-metric fitness function to enhance classification performance. PSO is employed to optimize the C parameter and RBF kernel of SVM, producing a more robust and balanced model compared to existing approaches. This study uses a heart disease dataset consisting of 1,025 rows with 13 attributes and 1 target variable obtained from the Kaggle repository and republished on the Zenodo platform in 2024. The research stages include Pre-Processing, Standardization, Feature Selection based on Correlation, and evaluation using the 10-Fold Cross Validation technique with Accuracy, precision, recall, and F1-score metrics. The results show that Support Vector Machine (SVM) achieved an Accuracy of 82.80%, Precision of 79.31%, Recall of 91.70%, and an F1-score of 84.88%. After optimization using PSO, the performance improved to an accuracy of 84.46%, precision of 80.54%, recall of 92.72%, and an F1-score of 86.04%. The experimental results indicate performance improvements of 2.00% in accuracy, 1.55% in precision, 1.11% in recall, and 1.37% in F1-score after PSO optimization. These results prove that the applied hybrid approach successfully improved the ability to detect heart disease. Therefore, this study contributes by demonstrating that PSO-based hyperparameter optimization can effectively enhance SVM classification performance for heart disease detection. The proposed model also has practical implications as a decision support tool for early heart disease detection that can assist medical practitioners in improving diagnostic accuracy and supporting preventive treatment strategies.]]>
Copyrights © 2026
