CVD Oriented HDPPP Categorization Method

摘要

CARTCM, a novel categorization method based on classification and regression tree (CART) is proposed to group 32 hemodynamic parameters (HDPs) and 5 physiological parameters (PPs) for detecting cardiovascular diseases (CVDs) efficiently. The CARTCM has two pivotal procedures: (ⅰ) compute importance of each HDP or PP variable automatically using CART and (ⅱ) depends on variable's importance threshold, categorize HDPs and PPs into 3 different groups (most, less and least import groups) on-line dynamically according to pre-detected type of CVD. In another word, for detecting different CVDs, the number and type of HDP&PP included in each of 3 groups are various. Solving such a refractory bottleneck problem improves greatly the accuracy and efficiency of detecting CVDs.