Model-Based Learning Techniques for Accurate Heart Disease Risk Prediction

The escalating global health crisis posed by cardiovascular disease, which claims an astonishing 17.9 million lives annually, necessitates precise and timely heart disease prediction, particularly in regions with limited access to advanced medical technology. Machine learning techniques have demonstrated potential in this regard, with previous studies achieving accuracies ranging from 86.6% to 90.78%. This paper addresses the urgency of accurate heart disease risk prediction by employing model-based learning techniques, focusing on four key algorithms: Support Vector Machines, Naive Bayes, Decision Trees, and Random Forests. Leveraging the Heart Failure Prediction dataset, which we split into training and testing sets, we systematically evaluated the performance of these models. Our findings reveal that Support Vector Machine achieved an impressive accuracy of 90.76%, emphasizing its proficiency in classification. Gaussian Naive Bayes displayed a balanced performance with an accuracy of 88.59%, maintaining an equilibrium between precision and recall. Although Decision Tree exhibited slightly lower accuracy at 82.07%, it maintained a well-balanced trade-off. However, Random Forest stood out as the top performer, with an accuracy of 91.85%, excelling in classification, precision, and true positive detection. These results provide crucial guidance for selecting an optimal model, and future research avenues include hyperparameter fine-tuning, ensemble methods, larger datasets, and the interpretability of model predictions, particularly in the medical field, where accuracy and transparency are paramount. Overall, this study underscores the potential of model-based learning techniques in accurate heart disease prediction, with Random Forest emerging as a promising solution.