[2K3-NFC-3a] [NFC Session] NFC-3 PADSED (Product and Activity Design Support based on Event Database)

[2K3-NFC-3a-05] Treatment strategy of Septicemia based on DPC data analysis by using pLSA method

〇Kazuya Yamashita¹, Yuichiro Sakamoto², Eiichi Sakurai^1,2, Yoichi Motomura^1,2 (1. National Institute of Advanced Industrial Science and Technology Artificial Intelligence Research Center, 2. Saga University)

Keywords:pLSA, Bayesian network, medical, DPC data

This paper shows our study about treatment strategy of septicemia based on DPC data by using pLSA, probabilistic latent semantic analysis, which find hidden segments in data. Our aim of using this method is modeling patients’ change along with time. In this study, we estimates segments from the data that consists of times when each patients had some medical treatment and name of treatments. We compared each segments from the points of death rate, medical cost and patients’ hospitalization. We analyze patients’ movement pattern between segments. We also pay attention to that these movements is related to medical decision.