An efficient control variate method for appointment scheduling with patient unpunctuality

Abstract

We consider the problem of appointment scheduling in the presence of unpunctual patients. To address this problem, many studies combine simulation with some heuristic search method. Such optimization techniques usually involve a huge number of repeated schedule evaluations which makes the simulation run-time an important factor. In this paper, we propose a variance reduction technique to reduce the computational work required to evaluate the performance of a certain schedule. The proposed method improves the precision of the simulation estimates through the judicious use of control variates and represents considerable improvement compared to traditional Monte Carlo methods. The evaluation approach is then included in a simulation optimization framework to provide insights into scheduling with unpunctual patients. We examine various unpunctuality distributions and queueing disciplines and show that the optimal scheduling policy may change when unpunctuality is considered. In many cases, scheduling patients in blocks of two or three patients better mitigates the negative effects of patient unpunctuality.