Healthcare resource location-allocation model incorporating consumer choice and its uncertainty

Abstract

Medically under-served areas (MUAs) are areas that have inadequate health resources due to non-availability or poor quality of provided care. The best way to deal with MUAs is to provide new care facilities considering users’ choice. Users’ choice depends on different resources of facilities and their preferences can change over time hence, they can choose facilities other than the provided ones to use. The goal of this study is to increase the usage of provided facilities by determining resources and locations of facilities at the same time. For this purpose, a recent location-allocation model, that uses the latent class model and almost robust optimization (ARO) for preference uncertainty, was used as the base model. Non-linear terms are generated when the resources that affect the choice becomes a decision variable. It is linearized into two stages of (1) piecewise linearization and (2) reformulation-linearization technique. If the choice factor is an integer, the optimal solution can be obtained by a commercial LP solver rather than an approximate solution. As a result of applying the model to MUAs about obstetrics care units in Korea, it was found that more MUAs can be covered by endogenizing the choice attribute variables.