An online algorithm for integrated scheduling of pre-treatment and treatment appointments in radiotherapy using deep reinforcement learning

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Not reported in abstract

• Primary Outcome: The DRL-based integrated scheduling algorithm outperforms benchmark online heuristic scheduling methods across experimental scenarios — specific metrics not reported in abstract

• Nnt Or Sensitivity: Not applicable in traditional clinical sense; no NNT, sensitivity, specificity, or hazard ratio reported. Scheduling performance metrics (e.g., mean waiting time reduction, percentage of patients scheduled within target timeframes) are implied but not quantified in available text

• Confidence Interval: Not reported

Clinical Application

Implementation feasibility is currently theoretical. Deployment would require: integration with existing Oncology Information Systems (e.g., ARIA, Mosaiq); prospective validation in a real radiotherapy department; clinical governance approval; staff training; and ongoing model monitoring. The online (real-time) nature of the algorithm is operationally attractive but technically demanding. No implementation roadmap or pilot data are presented. Australian radiotherapy services — delivered through both public hospital networks (e.g., Peter MacCallum Cancer Centre, major state cancer services) and private providers — face well-documented capacity pressures and waiting time challenges. The Royal Australian and New Zealand College of Radiologists (RANZCR) publishes benchmarks for radiotherapy waiting times, and the Australian Institute of Health and Welfare (AIHW) monitors cancer treatment timeliness. An integrated scheduling tool of this type could theoretically support compliance with RANZCR waiting time targets (e.g., radical intent treatment within 4 weeks of decision to treat). However, TGA regulatory classification of clinical decision-support software, and compliance with the Australian Digital Health Agency's interoperability standards, would need to be addressed before clinical deployment. PBS implications are indirect — improved scheduling efficiency could reduce the burden of treatment delays that necessitate additional supportive care. RACGP involvement is limited given this is a specialist oncology workflow tool. Patients with cancer requiring radiotherapy in departments operating multi-step pre-treatment pathways (simulation CT, dosimetry planning, oncologist review) followed by linear accelerator treatment courses. Most applicable to high-volume radiotherapy centres managing complex scheduling across multiple resource types.

Abstract

Scheduling patients for radiotherapy within the oncology department is vital for ensuring timely and effective treatment. The patient's therapy consists of a sequence of pre-treatment consultations, i.e. preliminary investigations to define the required treatment, followed by the radiotherapy treatment sessions on linear accelerators. Timely scheduling of pre-treatment consultations is crucial to prevent unnecessary delays in radiotherapy treatment, thereby impacting patient health. Therefore, we suggest integrating both scheduling problems. We employ a Reinforcement Learning-based scheduling technique to solve the multi-appointment scheduling problem in an online fashion. We design a reward function comprising three components to guide the agent towards learning an effective scheduling policy involving multiple appointments. Additionally, we compare our proposed model with benchmark online heuristics. We further provide insight into the reward function design, which steers the agent to learn a feasible policy that leads to high-quality solutions. Experiments show that our algorithm outperforms the benchmark heuristics.

References

1. 1.Frankish, M., & Maenhout, B. (2026). An online algorithm for integrated scheduling of pre-treatment and treatment appointments in radiotherapy using deep reinforcement learning. Health Care Management Science. https://doi.org/10.1016/j.asoc.2023.110436