A hybrid optimization algorithm for surgeries scheduling

This paper deals with the Operating Room (OR) planning problem at an operational planning level. The problem addressed consists of two interrelated sub-problems usually referred to as "advance scheduling" and "allocation scheduling". In the first sub-problem, the decisions considered are the assignment of a surgery date and an OR block to a set of patients to be operated on over a given planning horizon. The second aims at determining the sequence of selected patients in each OR and day. We assume that the duration of surgeries are random variables with known probability distributions. For each sub-problem an integer linear stochastic formulation is given. A hybrid two-phase optimization algorithm which exploits the potentiality of neighborhood search techniques combined with Monte Carlo simulation is developed to solve the overall problem. The approach developed searches for a feasible and robust solution designed to balance the trade-off arising between the hospital and patient perspectives, i.e. maximizing the OR utilization and minimizing the number of patient cancellations. The contribution of this paper is twofold. The former, more methodological, is to provide an efficient algorithmic framework to solve the joint advance and allocation scheduling problem taking into account the inherent uncertainty of surgery durations. The latter, more practical, is to provide a tool to develop robust offline OR schedules which consider the trade-off between reducing surgery cancellations and postponements while maximizing the operating theater utilization. To evaluate the efficiency of the proposed algorithmic approach, in terms of quality of solutions and solution time, we provide a computational analysis on a set of instances based on real data.