Dynamic appointment scheduling with forecasting and priority-specific access time service level standards

This paper analyzed a multi-priority system with access time service level requirements for dynamic arriving requests. Motivated by the public outpatient appointment systems, a dynamic scheduling algorithm is proposed for systems with demand influenced by seasonal and trend effects. Performances are measured by the proportion of requests that can access the service within the target time of their priority class, the access time percentiles and capacity utilization. The complexities of this problem include the problem size, time-varying demand with seasonality and trend, demand exceeding supply with no rejection of requests and the planning horizon considered. A goal programming model is formulated for the deterministic problem with service level objective assuming perfect information or forecasts. From the optimal properties, a simple scheduling rule with local exchange is developed. To forecast demand with possible updating during the rolling horizon, a forecasting approach incorporating linear trend for annual demand and logistic regression for patient class proportion is adopted. It is integrated with the scheduling heuristic in simulation to develop the dynamic access time rules. In the experiments designed based on three high-demand specialist outpatient clinics, demand scenarios are created to mimic different seasonal effects in a one-year horizon. Results showed improved access times for semi-urgent class and stable class patients with the largest 10–50% access times over a set of reported statistics.