CAUSAL MACHINE LEARNING APPROACH FOR CARDIOVASCULAR DISEASES DETECTION

Cardiovascular Disease is critical to health if not managed properly. Without appropriate treatment, cardiovascular disease could lead to complications such as heart disease heart attacks or strokes. WHO identifies cardiovascular disease as the leading cause of death worldwide which necessitates an accurate and prompt diagnosis for treating it. Given different methods in mitigating the problem, different clinical strategies for diagnosing CVDs are on the rise. Today outstanding methods such as the use of the electrocardiogram (EKG) and echocardiography (ECG), for cardiovascular disease diagnosis and treatment are being incorporated by clinicians. Even with this numerous methods health practitioners have tried to outline several strategies in mitigating cardiovascular disease, but a lot of people still finds it very difficult to control the risk factors (behavioral and physiological) that can cause CVD or are completely oblivious to the fact that they are at high risk. It is at this point that the need for scientific and medical reasoning to be resourceful, requires application of causal knowledge. Furthermore, this research adopts a Causal Machine Learning approach for cardiovascular disease detection. To achieve this, this research first performs the acquisition of hypertension data from Kaggle data repository, second, preprocessing of the dataset was conducted through normalization and subset attributes selection. Thirdly, Bayesian Network and Counter Factual Machine Learning modules were developed in order to detect the factual and counter factual percentages of each patient cardiovascular complications level. The Bayesian network module classification outlines the risk ratio of each symptom with regards to the present heart condition, while the counterfactual module output displays the contrary report from the actual Bayesian classification. Hence by examining the outcome of each of the Machine Learning modules, it can be deduced that the system could diagnose and predict heart disease effectively.