Research On Algorithms For Patient Scheduling In Hemodialysis Service Center

Patient scheduling in hemodialysis service center is the key of make full use of dialysis resources.It is difficult to solve large-scale problems effectively by several kinds of methods proposed in existing research.Therefore,it is necessary to develop more efficient optimization methods to solve the large-scale scheduling problem of hemodialysis service center under the weighted single-objective.On the other hand,this problem is also a manyobjective optimization problem,and it is necessary to design the high-dimensional multiobjective optimization method to solve the problem under the many-objectives.In the case of weighted single-objective,the column generation of patient scheduling in hemodialysis center is designed.Two decomposition methods of column generation in patient scheduling problem are compared,and the advantages of decomposition based on patient needs are illustrated.The solution process of column generation is given.Based on the practical scheduling case of hemodialysis center in Wuhan PA hospital,a comparative experiment is carried out among column generation,rollout algorithm and heuristic algorithm.The experimental results show that the column generation is effective,and compared with the other two algorithms,the scheduling results of column generation are more in line with the hemodialysis patients’ preferences,and its sensitivity to the scale of hemodialysis patients is much higher than the scale of dialysis equipment.In the case of many-objectives,a clustering-based high-dimensional many-objective evolutionary algorithm(CBEA)is proposed,and three selection strategies are designed based on fast non-dominated sorting,clustering method and dominated region().In the selection process,the non-inferior solution sets of each level are distinguished by fast nondominated sorting,and then the individuals of the same level are selected by clustering and .The comparative experiments of CBEA and NSGA-II under three alternative strategies are carried out.The experimental results show that CBEA-S3 can get the scheduling result which is close to the optimal solution set within the acceptable running time.