Optimizing Machine Learning Algorithms for Heart Disease Prediction in Healthcare: A Comparative Study
S. Subha, C. Balakrishnan, T. P. Anish, E Gokila, M. Nalini, R. Sıva Subramanıan
Abstract
The healthcare business faces significant challenges in successfully maintaining and evaluating large amounts of patient data. Data analytics has emerged as a viable tool for issue resolution and decision-making as a result of technological advancements in healthcare. This automated procedure improves the resilience of healthcare systems by collecting and analyzing data to provide more effective and cost-effective treatments. One critical use of technology, primarily machine learning, involves the identification and diagnosis of diseases, notably heart disease, which contributes significantly to global mortality. This research study analyzes how machine learning methods (NB, RF, SVM, and LR) can predict the existence of heart disease. Given that nearly 90% of cardiac diseases are avoidable, using machine learning for prediction is critical. This research study assesses the performance of different algorithms using performance metrics such as accuracy, precision, area under the curve (AUC), and F1-Score. The experimental results reveal Random Forest (RF) as the most reliable predictor of heart disease with an accuracy of 83.52% compared to other supervised machine learning algorithms. The Random Forest classifier achieves F1-Score, AUC, and accuracy values of 84.21 %, 88.24%, and 88.89%, respectively. This study highlights the potential of ML, notably the Random Forest algorithm, in improving the predictive capacity of healthcare systems for more effective heart disease prevention and treatment.