FO-RS-3TM: A High-Performance Ensemble Model for Heart Disease Prediction with Feature Selection and Hyperparameter Tuning

Despite huge advances in cardiac disease prediction, there are still critical research gaps. These include creating more accurate and reliable prediction models that can handle various data kinds and account for complex risk factor connections. A novel heart disease prediction method is presented in this study to cover these shortcomings. The dataset imbalance is addressed by using SMOTE. To manage outliers, Z-Score normalization is used. Firefly Optimization improves relevant feature selection. Random Search also improves 3-tier ensemble model performance by fine-tuning hyperparameters. The recommended strategy improves predictions by fixing dataset errors and modifying model parameters. As proposed, the FO-RS-3-tier Ensemble Model performs well with 98.53% accuracy. Also, accuracy, recall, F1 score, and AUC score are considerably high. This research improves heart disease prediction using cutting-edge methods, highlighting the need for enabling a comprehensive predictive analytics approach in healthcare. These findings suggest monitoring cardiovascular health and enhancing predictive modeling to improve patient outcomes.