A Systematic Review of Machine Learning, Deep Learning, and Explainable AI Approaches for Cardiac Disease Prediction

The cardiovascular (Cardiac) disease (CVD) is another factor that causes death among the global population most, and this is the reason why there is a high necessity to implement proper, effective, and interpretive diagnostic systems. The usage of machine learning (ML), deep learning (DL) and hybrid algorithms of artificial intelligence to predict heart diseases has enjoyed a widespread use in the recent years with different levels of success. The current paper provides a systematized review of the latest ML-, DL-, and hybrid-based systems to predict heart disease that include ensemble model, deep learning model, explainable artificial intelligence (XAI), and privacy-preserving model. The analyzed studies are evaluated based on datasets, classifiers, validation methods, data balancing and evaluation measures methods. It is illustrated in the analysis that the traditional ML models are most likely to have the prediction accuracy in between 80-90, the ensemble and hybrid models will most likely have the prediction accuracy in between 90-98. Recent explainable and fine-tuned ensemble methods show an accuracy of a level of 99, however, frequently under controlled conditions. Irrespective of the developments made, the issues include low interpretability, excessive benchmark data, imbalance in classes, and computational complexity. According to the research gaps that are mentioned, the given review reflects the necessity of clarifiable hybrid models that include high performing ensemble models like XGBoost and Deep learning-based feature fusion to increase the predictive quality, transparency and clinical usability.