On an Approach to Evaluation of Health Care Programme by Markov Decision Model

Abstract

In this paper, we consider the evaluation of periodic screening programme for woman breast cancer and formulate the model as a partially observable Markov decision process (POMDP). We convert a POMDP with finite state, observation state and action spaces to an equivalent completely observable MDP with continuous state and finite action spaces. By this approach, we have an optimal policy from dynamic programming (DP) equation in an equivalent MDP, but we focus on considering the evaluation in scenarios of periodic screening for participants with silent condition of breast cancer and seeking an answer which programme is better than others for themselves. The aim of this paper is, by using the data sets based on cancer registration and estimated parameters of survival rates and other ratios related to screening and diagnoses in Japan, to evaluate some scenarios of breast cancer screening programme in POMDP.