A well-known problem in healthcare operations research is the nurse scheduling problem, where the goal is to assign patient shifts to nurses in a way that respects various constraints (e.g. each nurse’s expertise, day-offs, etc.) Inspired by conversations with staff from the Beth Israel Deaconess Medical Center in Boston, we consider a related problem of scheduling MRI scans, a limited healthcare resource, for patients in a way that maximizes resource utilization and minimizes wait times.
We frame this problem as a dynamic constraint satisfaction problem, a variant of CSPs where constraints can be introduced or removed as time passes. We explore some existing algorithms for doing so and write a program that adapts these algorithms to our MRI scheduling problem. Our report can be found here; accompanying code can be found here.