A Comprehensive Analysis & Performance Evaluation of Machine Learning Techniques in Heart Disease Diagnosis

Heart attacks and other sudden deadly coronary artery diseases are still a threat to people all over the world, so it's important to diagnose them as soon as possible. The paper presents a systematic review of machine learning techniques in the assessment of heart attacks with a focus on the significance of prompt and precise diagnosis. We employed a methodologically agnostic approach to our research question, but we focused on three main areas of machine learning (ML) evolved over past ten years: supervised and unsupervised learning approaches, as well as deep learning applications on physiological, biochemical, and clinical data. A comprehensive analysis of models' performance, datasets, and clinical relevance was holistically conducted for 127 papers. Consequently, the review underscores the effectiveness of deep learning and ensemble methods, particularly when addressing large, multimodal datasets. Random forests and gradient boosting algorithms demonstrated superior accuracy (85-90<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>) in predicting the likelihood of a heart attack, while CNN's algorithms demonstrated a sensitivity of 98% in the detailed analysis of ECG signals.