Research On Multi-objective Nurse Rostering Problem

Nurses as an important part of the medical system,the quality of their work affects the patient’s recovery and the hospital’s medical level directly.However,in real life,nurses in our country are faced with the problem of labor intensity,high degree of job burnout,low job satisfaction and serious brain drain.A good nurse rostering model plays an important role in improving nurses’ satisfaction and reducing the turnover rate.Nurse rostering was studied earlier in foreign countries.At present,most of the research is to study the problem of single-objective nurse rostering.But in reality,the nurse rostering has to balance the interests of all parties,so it is a multi-objective model.Moreover,there is a large gap between the oversea and domestic rostering models and constraint requirements,so it is not suitable to directly for China’s hospital nursing condition to copy foreign nurse rostering model.Therefore,we needs to establish a multi-objective nurse rostering model based on our own national conditions.In operations research,nurse rostering is generally considered as a a combinatorial optimization problem.This paper is mainly divided into three parts: model building,algorithm solving and empirical research.First of all,this paper based on the medical practice in our country.The objective is reducing violating regulations of hospital,satisfying nurses’ personal preferences and assuring the average time distribution of nurses.This paper establishes a multi-objective nurse rostering model with giving full consideration to the soft and hard constraints and setting the time parameters,covering parameters and preference parameters.Secondly,according to the characteristics of multi-objective nurse rostering model,this paper designs a Genetic—particle swarm optimization algorithm of multilayer-coded.In this paper,we use multilayer-coded to encode the particles.The first layer indicates the nurse’s serial number and working date.The second layer indicates the working status of the individual,ie whether the nurse is working or resting.After that,a series of operations are carried out between particles,including individual crossover,group crossover,variation,screening of individual optimal solution set,screening group optimal solution set,etc.Finally,taking a nurse rostering in an orthopedic department of a hospital as anexample,this paper uses the developed model and algorithm by MATLAB software to solve and analyze the problem.It finds the change of the objective of the model in the non-inferior solution sets,compares the algorithm rostering with the actual rostering,and proves the superiority of the new model and the algorithm.