Quantitative Evaluation of Medical Consultation Recommendation based on the Transition Probability Model

Abstract

This paper proposes a new method that quantatively measures the effect of medical consultation recommendation for the diabetes. It uses a new transition probability model. The model is composed of 7 kinds of status related to the diabetes. The status is "no consultation", "consultation", "medication", "insulin", "dialysis", "healthy", and "secession". The method estimates the transition probability between status by analyzing both the medical examination data and the medical receipt data. It evaluates behavioral modifications of object members which are candidates participating in the program of medical consultation recommendation. The transition probability is adjusted by their behavioral modifications. This paper applies the method to the data managed by Toshiba health insurance union. Also, it verifies the effect of the method based on the simulated result.