An Adaptive Large Neighbourhood Search Procedure Applied to the Dynamic Patient Admission Scheduling Problem

The Patient Admission Scheduling problem involves assigning a set of patients to hospital beds over a given time horizon in such a way that several quality measures reflecting patient comfort, treatment efficiency, and hospital utilization are maximized. Usually it is assumed that all information regarding each patient is known in advance, making it possible to solve a static, offline planning problem. Such an approach, however, often has shortcomings in practice given the dynamic setting in which hospitals operate. An extension of this problem, known as the Dynamic patient Admission Scheduling problem, better reflects reality by attempting to capture, among other things, uncertainty in the length of patient stays as well as the ability to consider emergency patients. In this paper we devise an Adaptive Large Neighbourhood Search procedure, utilizing a Simulated Annealing framework, for this new variant of the problem and test its performance on a set of 450 publicly available problem instances of different size. A comparison with the current state-of-the-art indicates that the proposed methodology provides solutions that are of comparable quality for small and medium sized instances, but in a much shorter time frame. For larger instances the improvement in solution quality is dramatic, approximately 3-14% on average. In such cases, it does, however, take slightly longer.