3S-1
On Causal Relation Mining from Care Records
○Ahmad Zufri，蜂谷和士，Antonia Scheidel，橋本和夫（東北大）
It is important to analyze nursing records together with incident reports, called HIYARI-HATTO reports, to detect possible risks in the daily life of the elderly. However, it is difficult
for service providers to analyze such records in addition to the daily burden of elderly care.
In the hope of facilitating such analysis, the authors are applying topic detection technology
to nursing records to enable efficient discovery of health status changes.
This paper compares two approaches of causal relation mining, hidden Markov and
association rule mining, and discusses the feasibility to apply for the health care data
analysis.