Accommodating individual preferences in nurse scheduling via auctions and optimization

This paper describes a two-stage approach to nurse scheduling that considers both nurse preferences and hospital constraints. In the auction stage, nurses bid for their preferred working shifts and rest days using "points". An optimization model awards shifts to the highest bidders insofar as possible while maintaining hospital requirements. In the schedule completion stage, an optimization model allocates the unfilled shifts to nurses who have not yet met their minimum required hours. The approach is demonstrated via a case study in the emergency department at York Hospital. A schedule with a high percentage of awarded bids was generated in a few minutes of computer time. Further experimentation suggests that the approach works well under a variety of conditions.