Hospital Surgery Scheduling Under Uncertainty Using Multiobjective Evolutionary Algorithms

Abstract

Surgery scheduling is the allocating of hospital resources to surgical procedures over time. These problems are continuously encountered in hospitals delivering surgical treatment to patients, and they must be solved with great effort and care in order to utilize scarce resources, balance conflicting interests and hedge for uncertainty. Therefore, machine learning algorithms are generally not directly applicable to surgery scheduling problems. The motivation of this work is to narrow the gap between evolutionary approaches to machine scheduling problems from literature and their practical applications to real-world hospital surgery scheduling problems. We formulate a new variation of the surgery admission planning problem and develop evolutionary mechanisms to solve it with state of the art multiobjective evolutionary algorithms. Our most important contribution is a fast simulation approach for evaluating surgery schedules under surgery duration uncertainty. By using Monte-Carlo simulation to estimate the fitness of individuals during optimization, we attempt to find solutions that are robust against variations in surgery durations.