Abstract
In this research, healthcare software for multiple people is studied. Healthcare software gives appropriate advices based on personal health states. In previous research, probability of staying in objective states is maximized for one person. This research targets two problem settings and two rewards. In the first problem setting, one common advice is selected for multiple managed people. In the second problem setting, one advice is selected for each managed person. The first reward represents the sum of the probabilities that each managed person stays in the objective states. The second reward represents the probability that all managed people stay in the objective states at the same time. A new advice selection method is proposed for each couple of problem setting and reward. The proposed methods maximize expected rewards based on statistical decision theory by dynamic programming. The effectiveness of the proposed methods is shown by some computational examples. In the computational results, adaptive advice selection examples for each couple of problem setting and reward are confirmed. This research is basic research, and future extended research is required.
