Profiling Cardiovascular Disease Event Risk through Clustering of Classification Association Rules

Abstract

Association Rule Mining (ARM) is a promising method to provide insights for better management of chronic diseases. However, ARM tends to give an overwhelming number of rules, leading to the long-standing problem of identifying the 'interesting' rules for knowledge discovery. Therefore, this paper proposes a hybrid clustering-ARM approach to gain insight into a population's pattern of risk for a chronic disease related adverse event. Classification Association Rules (CARs) indicative of the development of cardiovascular disease (CVD) are developed from training data and clustered based on commonality of cases satisfying the rule antecedents. Test cases are then assigned to the rule clusters to provide sets of at-risk individuals sharing common CVD risk factors. The approach is demonstrated using the Framingham Heart Study cohort data set obtained from the US National Heart, Lung, and Blood Institute's Biologic Specimen and Data Repository Information Coordinating Center (BioLINCC).