Heart failure remains a major global cause of death, making accurate mortality prediction crucial for effective clinical decision-making. This study proposes a mortality classification model using a Support Vector Machine (SVM) optimized with the Firefly Algorithm (FA). This approach overcomes the limitations of conventional machine learning methods that rely on manual hyperparameter tuning and frequently produce suboptimal performance when applied to complex clinical datasets. The dataset contains 299 heart failure patient records with 12 clinical attributes and one target label. Preprocessing steps include winsorizing for outlier, feature standardization, and handling class imbalance using the Synthetic Minority Oversampling Technique (SMOTE). FA is applied to automatically optimize the SVM hyperparameters, specifically cost (C) and gamma (γ), to improve classification performance. The proposed FA–SVM model achieved 90.24% test accuracy and a recall score of 0.93 for deceased patients, demonstrating strong capability in detecting high-risk cases. These results indicate that combining FA with SVM improves predictive accuracy and model stability, showing promise as a clinical decision support tool for early heart failure risk detection.
Khoirunnisa et al. (Tue,) studied this question.