A Dynamic Nurse Scheduling Approach Using Reinforcement Learning to Address Sudden Absences of an Unknown Nurse

Abstract

Creating work shift schedules for nurses can be a complex task, as it involves satisfying various requirements that can be difficult to reconcile. Although several studies have investigated the nurse scheduling problem, creating practical work schedules with numerous constraints and evaluation values can still be challenging. To address this issue, we have proposed a method for work revision that utilizes reinforcement learning to improve a constructive nurse scheduling system. In this article, we extend the proposed method to accommodate dynamic nurse scheduling, wherein work schedules are revised or rescheduled in response to sudden absences. Specifically, we demonstrate the effectiveness of our approach in creating feasible work schedules for an unknown nurse who may be absent at any given time, through computational experiments.