Research On Gynocological Ward Bed Scheduling Based On Mixed Integer Programming

As one of the key resources in a hospital, the management ofhospital beds has always been paid much attention. With limited numberof beds, how to utilize the operational research method and theory tomake appropriate admission scheduling for awaiting hospitalized patientshas already been a hot research topic within the healthcare managementfield. Nowadays, most hospitals still make the bed allocation decisionmanually following the FCFS (First Come First Service) rule because ofthe technological limitation. Under these circumstances, this paper willstudy the problem of hospital bed planning based on the on-site survey ofa real gynecological ward in Jiangsu Province. Through the survey wehave not only understand the hospitalization process of a gynecologicalward but also collected2012whole year operational data, which hassubstantially support our study. In this paper, we will differentiate thehospitalization patients into three categories according to their utilizationof ward resources (beds and surgeons), namely emergent patient, normalsurgical patient and normal non-surgical patient.More specifically, this paper discussed a problem under followingassumptions: there is a list of awaiting hospitalized patient and their basicinformation are confirmed, including preferred hospitalization period,disease category, daily delay cost, daily revenue and LOS(length of stay,or LOS distribution); Within the planning period, the status of all bedsand daily surgery capacity are all known. Upon these, we are trying tofind out a method for working out a reasonable plan for hospitalized patients waiting for admission arrangements, which is to determine whichpatients can be admitted in the planning period, and to determine thepatient’s first hospitalization day and the bed number she will stay in tominimize total bed arrangement cost.Based on the above issues, this paper first considers a deterministichospitalization bed arrangement mixed integer programming model, inwhich assumes that the hospitalization days are determined and known.Application of standard linear programming solver CPLEX helps to solvethe model, and the results were compared with FCFS rule arrangementresults, which confirms that the proposed model can reduce patient totalwaiting and transfer cost. Then, considering the reality that patient’s LOSis often uncertain, this paper further considers random LOS and modifiedthe deterministic LOS bed arrangement model accordingly. In thestochastic LOS model, we no longer assumed that the patient’s LOS wasdetermined, but used discrete distributions of LOS which is obtainedaccording to the patient’s symptoms, diagnosis and data of the historicaltreatment, to characterize the hospitalization period. Through a series ofnumerical experiments, we used both of the proposed method for workingout the arrangement of same group patients, and compared the totalstochastic cost of both arrangement results. The result showed that thestochastic LOS bed arrangement model can obtain better arrangementplan.