Using Optimization Techniques and Multidisciplinary Collaboration to Solve a Challenging Real-World Residency Scheduling Problem

In this paper, we consider the problem of assigning medical residents to shifts within a pediatric emergency department. This problem is challenging to solve for a number of reasons. First, as with many other healthcare personnel scheduling problems, it has a nonhomogeneous workforce comprising residents of different characteristics, requirements, and capabilities. Second, residency scheduling problems not only must ensure adequate resources for patient care but also must meet educational training needs, adding further complexity and constraints. Finally, rather than being evaluated under a single cost metric, resident scheduling problems have multiple objective criteria, which are often in conflict with each other. Some of these challenges can be overcome through the use of operations research techniques; others depend on the process by which we apply these techniques and, in particular, the way that the operations researchers collaborate with the clinicians. We present our experience at the University of Michigan C.S. Mott Children’s Hospital in building monthly schedules, focusing on both our integer programming formulation and the iterative, interactive approach in which we use this integer program as a tool within the broader process of schedule development.