A Study of Nurse Rostering Problems

Most hospitals operate 24 hours a day, 7 days a week and face widely fluctuation demand. A more suitable arrangement for working hours and days off is needed, especially in the view of the growing nursing shortage. Therefore, nurse rostering problem (NRP) is an important issue in the most of hospitals. In the NRP, the legal requirements, the hospital objectives and nurses’ preference should be taken into consideration. Legal requirements includes government’s law and contractual obligations; hospital objectives may include ensuring a continuous service with required nursing skills and the number of nurses, while avoiding additional costs for unnecessary overtime. Nurse’ preference includes fairness considerations, such as the ratio of night shifts or day off for a roster, and the desirability (undesirability) of certain shifts of scheduling patterns. NRP is a very time-consuming task of constructing schedules that satisfy both the requirements of hospital and the preferences of personnel is still performed manually in many cases. Therefore, how to generate an acceptable roster using reasonable computational time is one of issue to be solved in this proposal. A disruption in a roster is defined as an occurrence when a nurse is unavailable to work a planned task due to unplanned absences (i.e., nurse turnover). When a disruption occurs in the roster, the current roster is usually infeasibility. Hence, the scheduler must re-assign other nurses to cover the nurse such that the current nurse requirements and time related constraints are satisfied. Therefore, how to re-generate a new roster which must comply with the labor rules and institutional constraints, and the new roster must be as similar as possible to the current one quickly, is another important issue to be discussed in this proposal. NRP belongs to NP-hard problem; therefore, how to find a (near) optimal solution using reasonable computational time is a critical challenge. This proposal wants to develop an artificial immune system (AIS) algorithm and an artificial bee colony (ABC) algorithm to solve NRPs, and apply the proposed algorithms to solving the NRP in the emergency department of a medical center in the north Taiwan. This proposal plans to generate the reasonable roster schedule within the acceptable computational time and regenerate the roster scheduling instantly if something is change after the initial roster is obtained. In addition, the proposed approaches are planned to apply to problems in three public datasets, to show the ability to solve different kind of situation instead of only one specific hospital. A user-friendly interface will be provided to scheduler to solve the real problem without difficulty.

Project ID:PB10307-0472
External Project ID:MOST103-2410-H182-006