Authors: Su-Dong Lee and Chi-Hyuck Jun
Hypertension is one of the biggest burdens of public health all around the world. In hypertension prevention strategies, targeting individuals at high risk of developing hypertension is a key task. To build a hypertension risk prediction model, we obtained a cohort data set collected by Korean National Health Service. In order to analyze categorical-numeric mixed type data, we propose a novel approach named two-stage logistic model tree. Experimental results show that our proposed method is an effective approach to improving classification performance with mixed type data for hypertension risk prediction.
Keywords: hypertension; risk prediction; mixed type data