Coordinating oncologist appointments with chemotherapy treatments under uncertainty

Scheduling chemotherapy treatments in outpatient chemotherapy clinics (OCCs) presents significant challenges due to limited resources, uncertainty in infusion durations, and the need for coordinating with oncologist consultations. This study addresses these challenges through integrating oncologist consultation and chemotherapy scheduling by determining appointment times for a daily list of patients. A two-stage stochastic mixed-integer programming model is developed, considering stochastic factors such as infusion durations and the statuses of chemotherapy treatment approvals after consultations. In the first stage of the model, patients of each oncologist are organized in a sequence, and appointment times are set. In the second stage, patients are assigned to chairs and nurses using an optimal myopic policy. The objective function penalizes the expected weighted sum of the total working time of the OCC and the waiting times of patients. To represent the original scenario set by a reduced scenario set, a scenario reduction algorithm is employed. The algorithm, a Wasserstein Distance-Based Local Search Algorithm (WD-LSA), is tested using real data from a major academic oncology hospital in Turkey. The performance of the WD-LSA algorithm is demonstrated by comparing with CPLEX for smaller number of scenarios and with heuristic algorithms for larger number of scenarios. We find that the gap is quite small when compared with CPLEX solutions and the solutions are much better than the solutions found by the practical scheduling heuristics from the literature. The trade-off between patient waiting time and total working time is assessed. The dependency of the performance measures to the number of oncologists and nurses is investigated. Lastly, the value of stochastic solution is estimated.