Stochastic bi-level optimization models for efficient operating room planning

Within a hospital, the operating room (OR) department has the largest cost and revenue. Because of the aging population, the demand for surgical services has been increasing sharply in recent years. At the other hand, the rate of OR capacity expansion is lower than the rate of increasing demand. As a result, OR managers must leverage their resources by efficient OR planning. OR planning is challenging because of multiple competing\conflicting objectives such cost minimization and throughput maximization. Inherent uncertainty in the surgical procedures and patients arrivals complicate the decision making process. This increases the risk of non-realization of the system objectives. In this paper, stochastic bi-level optimization models were formulated to optimize total cost and throughput of ORs under the presence of uncertainties in patient arrivals and case times. Newsvendor model and chance-constrained optimization method were used to optimize multiple objectives under the presence of uncertainties. Using historical data, a simulation model was established to validate the results of optimization models. Using statistical process control (SPC) stability of each model was investigated. Using bi-level optimization, we addressed managerial preferences over total cost and throughput. Optimizing one objective may lead to compromise on the optimality of the other objective, which generates trade-offs. Using a trade-off balancing model, we found solutions that minimize the sum of deviations from the best solutions for both total cost and throughput. Trade-off balancing optimization models may lead to better solutions, compared to traditional multi-objective optimization models. The results of this paper are applicable to manufacturing systems, where managers face multiple objectives and uncertainties in the system. Published by Elsevier B.V.