Weighted Association Rules and Scoring Methodology for Cardiovascular Diseases

Abstract

Cardiovascular diseases (CVD), including coronary artery disease (CAD), myocardial infarction, and stroke are a group of highly prevalent and deadly diseases. The deaths from cardiovascular diseases were announced as 17.9 million in 2016 and it is expected that this number will reach approximately to 23.6 million by 2030. In order to facilitate the diagnosis and treatment of CVD, several computational approaches and data mining methods have been proposed until now. In this study, Apriori algorithm is utilized to find associations between features and rules based on UCI’s publicly available Cleveland dataset. Additionally, we generate different weighted association rules, which can help medical doctors to stratify patients and thus, propose different treatment approaches for each patient’s sub-category. Performance results show that the Apriori algorithm creates 58 rules when support and confidence parameters are set to 0.1 and 0.9, respectively. Utilizing weighted association rule approach, 6 important rules have been created based on Clinical Important Factors (CIF) and Framingham Heart Study Risk factors (FHS RF) on CVD.