PATIENT FLOW QUEUING MODEL: THE CASE OF ARBA MINCH GENERAL HOSPITAL OPDs

Abstract

Queuing is a major challenge for healthcare services all over the world, but particularly so in developing countries. Application of queuing theory to enhance decision making to improve this problem is not commonly used by managers in developing countries in contrast to their counterparts in the developed world. This study attempted to develop and analyze a patient flow queuing model in the case of Arba Minch General Hospital. To achieve this objective, primary sources of data was conducted from some selected service nodes of the hospital for construction days’ observation and obtained the values of parameters for each, arrival rates (λ) and service rates (μ). The required data and computation for the study was analyzed using a MATLAB computer programme and Microsoft Excel. The author formulated and analyzed a queuing network of patient inflow, rerouted several service nodes, and monitored the outflow of the healthcare system. The investigation of the study was analyzed the patient flow queuing model in AMGH and the steady-state (routing) probabilities was derived by Moore-Penrose or pseudo-inverse methods. Healthcare systems differ intrinsically from manufacturing systems. Thus, the hospital system requires a distinct modeling approach. In this thesis, the author showed how to construct a queueing network model consisting of several service nodes of Arba Minch General Hospital based on the diagram assumed was captured with the support of real data. In this study, the focus was on the level of service the customer needed rather than a faster server. In order to analyse such networks, the Moore-Penrose pseudo-inverse methods were used and obtained the routing probabilities of each service node. Based on the routing probabilities, effective arrival rates and performance measures of the hospital queuing network were calculated. The network was decomposed into a set of single queueing systems and analyzed separately. Therefore, results of the single queueing systems that aggregated and general performance measures (i.e., average number of patients in queue(𝐿𝑞) and in systems (𝐿), average time a patient spends in queue (𝑊𝑞) and in systems (𝑊), server utilization time (ρ), and idle time (𝜋) in each service node) of the queueing network were obtained. In addition, the author developed new constructing queuing network models to assess the impact of service outages, use the queueing network to approximate patient flow times, and evaluate a number of recommended servers. The result of performance measures has suggested a solution to accelerate the outpatient department services of an Arba Minch general hospital.