Heart Disease Prediction Evaluation of Machine Learning Models with PSO-Optimized K-Fold Cross-Validation

Heart disease prediction is a critical aspect of modern healthcare and demands accurate and efficient models for timely diagnosis. Machine learning models in heart disease prediction analyze patient data, with cross-validation ensuring reliable performance, aiding model selection, and optimizing parameters for enhanced accuracy. The k value in cross-validation determines the number of folds into which the dataset is divided for training and validation. In the proposed work, optimization of the k-value in cross-validation is done using Particle Swarm Optimization (PSO), which is utilized to improve the evaluation of the predictive accuracy of various models. PSO dynamically adjusts the k-value, balancing computational efficiency with accurate model evaluation. The result shows that among various ML models, Random Forest, with an 86.95% mean accuracy, emerges as the preferred choice, offering reliable predictions for improved diagnostics.