Robust Multi-Objective Operating Room Scheduling under Stochastic Emergency Arrivals

Abstract

Operating room scheduling is a central problem in healthcare resource management, and the inherent uncertainty in emergency patient arrival times poses a major challenge for scheduling decisions. This paper develops a robust multi-objective optimization framework integrated with Monte Carlo simulation, employing an integrated Non-dominated Sorting Genetic Algorithm II to address the single operating room scheduling problem. The proposed model simultaneously optimizes two competing objectives, namely weighted operating cost and maximum start-time delay, and introduces a risk-aversion coefficient γ to achieve a flexible trade-off between expected performance and robustness. Within this framework, emergency patient arrival times are modeled using a uniform distribution U(30, 180) min, while the planned start times of elective surgeries are perturbed within a delay interval of 0–40 minutes. Experimental results based on a 20-case surgical dataset demonstrate that the proposed algorithm generates six Pareto-optimal solutions, with robust weighted-cost values from 18,255.5 to 19,803.9 and robust tardiness values ranging from 591.8 to 615.3 minutes. A cost-tardiness trade-off analysis reveals that an 8.48% increase in the weighted cost is associated with a 3.82% reduction in maximum start-time delay, thereby providing hospital administrators with a quantitative basis for decision-making.