MDP based inpatient premature discharge decision making in hierarchical medical system

In Chinese hierarchical medical system, hospitals are divided into different levels according to different types of treatment. However, due to lack of clear regulation, many patients visit upper-level hospitals even for simple services while more severe patients may encounter denial of care. To solve this problem, reverse referral, i.e., transferring premature inpatients from upper-level hospitals (ULH) to lower-level hospitals (LLH) has been proposed. Therefore, how to make decisions for the crowded ULH to implement the reverse referral and at the same time maximizing its profit is quite essential to be studied. The classical dynamic programming method of Markov Decision Process (MDP) is applied to build the inpatient referral model in this paper. We provide an optimal dynamic control policy and all the numerical results are analyzed. The results show that an optimal referral policy is feasible in the real life and helpful to motivate reverse referral.