Radiotherapy treatment scheduling: Implementing operations research into clinical practice.

Abstract

RESULTS

The weekly schedule was improved in both centers by decreasing the average standard deviation between sessions' starting times from 103.0 to 50.4 minutes (51%) in one center, and the number of gaps in the schedule from 18 to 5 (72%) in the other. The number of patients requiring linac switching between sessions has also decreased from 71 to 0 patients in one center, and from 43 to 2 in the other. The automated process required 5 minutes and 1.5 hours of computation time to find an optimal weekly patient schedule, respectively, as opposed to approximately 1.5 days when performed manually for both centers.

BACKGROUND

Every week, radiotherapy centers face the complex task of scheduling hundreds of treatment sessions amongst the available linear accelerators. With the increase in cancer patient numbers, manually creating a feasible and efficient schedule has shown to be a difficult, time-consuming task. Although operations research models have been increasingly reported upon to optimize patient care logistics, there is almost no scientific evidence of implementation in practice.

CONCLUSIONS

The practical application of a theoretical operations research model for radiotherapy treatment scheduling has provided radiotherapy planners a feasible, high-quality schedule in an automated way. Iterative model adaptations performed in small steps, early engagement of stakeholders, and constant communication proved to facilitate the implementation of operations research models into clinical practice.

METHODS

A mathematical operations research model was adapted to generate radiotherapy treatment schedules in two Dutch centers. The model was iteratively adjusted to fulfill the technical and medical constraints of each center until a valid model was attained. Patient data was collected for the planning horizon of one week, and the feasibility of the obtained schedules was verified by the staff of each center. The resulting optimized solutions are compared with the ones manually developed in practice.

More about this publication

PloS one
  • Volume 16
  • Issue nr. 2
  • Pages e0247428
  • Publication date 20-02-2021

This site uses cookies

This website uses cookies to ensure you get the best experience on our website.