Machine Learning Models for Heart Disease Prediction: Balancing Accuracy and Transparency

Heart disease continues to be the world's leading cause of death, so precise and trustworthy predictive models are required to enable early diagnosis and individualized treatment. This paper addresses the escalating global burden of heart failure, marked by increasing morbidity, mortality, and strain on healthcare systems, stemming from left ventricular systolic and diastolic dysfunction. When compared to conventional methods, machine learning proves to be a more accurate tool for predicting heart diseases. Machine learning (ML) models have shown great promise in predicting heart disease, but their complexity often results in a trade-off between accuracy and interpretability. This study investigates various interpretable machine learning approaches, including decision trees, Mtree, Random Forest, K-means, to predict heart disease, while maintaining transparency. Utilizing a dataset comprising patient health records, including attributes like age, blood pressure, cholesterol levels, and electrocardiogram results, preprocess the data through normalization and feature selection to enhance model accuracy. Our study showcases the implementation of distinct algorithms, demonstrating their effectiveness in accurately detecting heart failure with an impressive accuracy range of approximately 89% to 92%. The study underscores the importance of balancing accuracy with transparency in developing ML models for healthcare applications, particularly in high-stakes scenarios like heart disease prediction. Keywords: Heart disease, Machine learning (ML), random forest, health records, decision trees, MTree.