Predicting Ischemic Heart Disease Using Supervised Learning Techniques on Clinical Data

Abstract

Ischemic Heart Disease (IHD) continues to be one of the world's leading causes of death, accurate predictive models are fervently needed to enable early intervention. Traditional risk assessment methods often fall short in timely detection and prevention, necessitating the development of advanced machine learning (ML) models. This review critically evaluates recent advancements in ML-based IHD prediction, focusing on their objectives, performance metrics, limitations, and datasets. Our analysis identifies Random Forest and Support Vector Machines as the most frequently adopted models, achieving accuracies often exceeding 90%. However, challenges such as small, imbalanced datasets and data privacy issues persist. Emerging trends highlight the importance of comprehensive datasets and personalized approaches to enhance model reliability and clinical applicability. Additionally, there is a growing emphasis on feature selection and data quality to improve prediction accuracy. This research proposes RF model for accurately predicting IHD. The results show that the proposed model outperforms current works with 0.93 accuracy and 0.95 Area Under the Curve.