Multi-level Nurse Scheduling Problem With Multi-level Nursing Positions Based On Evolutionary Algorithm

Nurse scheduling plays an important role in hospital operations.An efficient schedule may affect considerably the operating cost and nursing quality of a hospital.Unfortunately,nurse schedules are compiled manually in China and the research is lacking in considering nurse scheduling with multi-level nurses and multi-level nursing positions.Considering the real-world routines and preferences in China,this paper proposes a goal programming model for the multi-level nurse scheduling problem with multi-level nursing positions.To solve the model,an evolutionary algorithm based nursing scheduling approach(ENS for short)is devised,which can improve the nurses' satisfaction and reduce the cost of hospital care.The main contents of this paper are as follows:(1)According to the real-world operations in hospitals,a nurse schedule is defined under consideration of nursing needs,labor regulations,nurses' preferences and nursing costs.Different from the traditional nurse scheduling problem,this paper considers characteristics of multi-level nurses and multi-level nursing positions of the problem,taking into account the matching of the two.(2)A mathematical programming model for the nurse scheduling problem is established.The objective is to minimize the nursing cost and maximize the satisfaction to nurse preferences.The constraints include those related to nursing demands and labor laws.This paper systematically analyzes and summarizes the common labor regulations and actual needs in hospital daily operations.Since many contraints related to labor laws and regulations exist,an initial feasible solution,as the starting point of searching process,is hard to be compiled.Therefore,this paper classifies the constraints into hard rules and soft rules.The model is improved by transferring the soft rules into the model objectives.(3)The ENS approach is proposed,which compiles first an initial solution based on nursing needs,then refines the solution using an evolutionary algorithm iteratively.Each iteration contains four operations upon a solution: decomposition,evolutionary destruction,randomly rebuild and acceptance check.The evolutionary process is carried out iteratively until the certain ending condition is met.In the stage of evolutionary destruction,the mechanism of the survival of the fittest is embedded to improve the solution quality.In the stage of randomly rebuild,four different scheduling strategies are proposed to further speed up the rebuild speed and ensure the quality of reconstruction.Experiments show that ENS can effectively compile schedules with lower cost and higher preference satisfaction than manual schedules.