One of the causes of death globally is cardiovascular disease (CVD). Early prediction techniques are an essential necessity to deal with this problem. The Machine Learning (ML) developed methods have enormous opportunities in the risk prediction of CVD. Enhancing accuracy, interpretability, and integration of various data sources are useful methods that can be used to predict early CVD. This literature review article will outline the more recent advancements in ML to predict CVD. These researches rely on the data retrieved from electronic health records, medical images, wearable equipment, and lifestyle and psychological aspects of patients. ML algorithms and techniques, including Random Forest, Gradient Boosting, Deep Neural Networks, and Hybrid Models, as well as Ensemble Models, have been utilized. These approaches improved performance and accuracy in terms of AUC benchmarks. Other researches even have mental health metrics to enhance prediction accuracy. A small generalizability across populations, meanings of imbalance in datasets, and the lack of standardized evaluation procedures and difficulties in clinical incorporation remain significant concerns. The survey provides a synthesis of research gaps and offers future directions on how to further develop robust external validation and integrate it into the real-time medical support systems to help it enter much more widespread use in healthcare practice.
Sivakumar et al. (Fri,) studied this question.