Handling time-related demands in the home care nurse-to-patient assignment problem with the implementor-adversarial approach

Improving accessibility to care in medically under-served areas(MUAs) is the goal of MUA support program in public health policy. In the planning phase of such programs, we often use geographical proximity of care facilities as a measure of accessibility, and the programs resource is used to maximize the geographical accessibility. While it is easy to assess the geographical accessibility, this is not always an accurate assessment of the actual, realized accessibility because true accessibility is realized by the actual service use by patients. The choice of a specific care provider by patients is made not just by a physical distance, but by many other factors including the size of a care provider, physician’s demographic, etc. Predicting true accessibility thus requires a model that considers various factors in the patients decision making, and in this paper, we use a choice model known as the conditional logit model. We use the actual health insurance data from Korea to identify factors affecting patients choice of care providers and model the provider choice behavior of patients by using the MNL model. To validate the proposed model, we compare the actual patient volumes for care providers with the model prediction, and the results show a good agreement suggesting the MNL model is a promising approach to assess true accessibility to care.