Early Detection of Cardiovascular Diseases

Cardiovascular diseases (CVD) represent a significantly perilous category of health conditions that directly affect the heart as well as the blood vessels. Notoriously recognised for their critical impact on global health, these diseases stand at the forefront among the primary reasons for mortality worldwide, notably accounting for the highest number of deaths attributed to non-communicable diseases. Given these alarming statistics, it becomes imperative to develop methods and create models capable of accurately predicting the onset of CVD among individuals who are currently perceived as healthy. This approach is not only essential for the early detection and subsequent management of such conditions but, most crucially, serves as a cornerstone for the prevention strategies aimed at mitigating the risk and potential severity of cardiovascular diseases for those who might fall ill in the future. This study proposes a promising machine learning-based approach for early CVD detection and is comparing various state-of-the-art techniques. The methodology is applied on the Framingham dataset aiming to indicate the possibility of developing a coronary heart disease (CHD) within ten years. With an accuracy of 93%, the stacking classifier model with synthetic data outperformed all existing approaches applied on the same dataset. The obtained results are indicating that approaches like the one we present hold great potential in revolutionising CVD detection.