Abstract
In this research, healthcare software for personal health management is studied. Healthcare software gives appropriate advice based on personal health data. In previous research, probability of staying in objective states is maximized under the condition that true health state is known. But it is not always possible to determine the true health state from the observed data. In this research, a new method which maximizes probability of staying in objective states under the condition that true health state is unknown is proposed. Markov decision processes with unknown state is used in the modeling. Dynamic programming is used in the proposed method. The effectiveness of the proposed method is shown by some computational examples. In the results, probability of staying in objective states by the proposed method is greater than that of the comparison target. This research is a basic research, and future extended research is required.
