Compatibility of short and long term objectives for dynamic patient admission scheduling

•A full Mixed Integer Programming (MIP) formulation is provided for the DPAS.•A short term objective design principle is introduced for dynamic scheduling.•A new state of the art short term strategy is developed for the DPAS.•Three MIP formulations are applied for solving short term DPAS problems.•Lower bounds are calculated using Dantzig-Wolfe decomposition and column generation. When applying periodic re-optimization for handling a dynamic scheduling problem, the objective of the problem solved in each period (its short term objective) significantly impacts the quality of final solutions (its long term solutions). Meanwhile, designing a short term objective consistent with the dynamic problem’s long term objective remains a very challenging problem in its own right. This paper studies the compatibility of short term and long term objectives in the context of the Dynamic Patient Admission Scheduling Problem (DPAS). A new short term strategy — which considers idle resource penalties and anticipatory information — is presented for the problem. The resulting approach is then applied to the available DPAS benchmark, with its long term solutions evaluated with respect to new lower bounds calculated using Dantzig–Wolfe decomposition and column generation. The results demonstrate that the proposed short term strategy produces long term solutions of significantly better quality than the best-known strategy for 26 out of the 30 instances.