A variable neighborhood search algorithm for the surgery tactical planning problem

We address the tactical planning problem of surgeries that consists in building an admission plan of patients over a medium-term horizon planning so as to minimize over and under utilization of several resources such as operating theaters, beds and nursing care, compared with their target level of utilization. The problem is formulated as a mixed integer linear program for which exact solution methods fail to find an optimal solution in a reasonable execution time. We develop a Variable Neighborhood Search algorithm and show its ability to provide high quality solutions in short computational running times compared with CPLEX for numerous real-sized instances based on the surgery planning problem in a Dutch cardiothoracic center. Furthermore, with few parameters' settings and low computational memory requirements, this approach may easily be implemented in a decision support system for hospitals. HighlightsWe consider the surgery tactical planning problem with multiple resources.We propose a Variable Neighborhood Search (VNS) approach to solve this problem.Simulation results show that VNS outperforms CPLEX in most cases.