Metaheuristics for the Multiobjective Surgery Admission Planning Problem

This paper presents a comparative survey on metaheuristics applied to the multiobjective surgery admission planning problem. Three well-known metaheuristics: genetic algorithm (GA), simulated annealing (SA) and variable neighbourhood descent (VND) are compared using the Wilcoxon signed rank test. The metaheuristics are also benchmarked against a hybrid GA that uses the VND as a local search procedure. The weighted sum method is used to balance five competing objectives: operating room overtime, operating room idle time, surgeon overtime, surgeon idle time and patient waiting time. As a preparation for future multiobjectivity analysis, a simple example shows how several non-dominated trade-off solutions may be presented to the decision maker using the $\epsilon$-constrained method and the non-dominated sorting genetic algorithm II (NSGA-II). The results are meant to serve as a starting point for further development and testing where the challenge of uncertain surgery durations will be included.