A Study of Neural Network Based Flexible Reservation Systems

• Main Author: 趙明彥
• Other Authors: Jy-Hsin Lin
• Format: Others
• Language: zh-TW
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/87738868368658785837

碩士 === 華梵大學 === 工業管理學系碩士班 === 90 === Although appointment systems have been used by the most clinics and hospitals to reduce patients’ waiting time, traditionally, appointment systems are designed from management view; these systems overbook patients to increase resources utilization, which cause long patients waiting time. Therefore how to design an appointment system not only from the management’s view but also from patients’ view to balance waiting time and resources utilization has become an unavoidable issue that must be addressed. In order to reduce patients’ waiting time and to maintain resources utilization, this study first analyzed the data collected from the hospital, and clustered patients with similar processing times. Next, for each individual environment, which consists of various clusters, the research developed a most feasible scheduling rule that yields the better performance. Finally the study used Backpropagation Neural Network to identify any significant changes in the environment, so that the appointment system could choose the appropriate scheduling rule according to the every of environment. The results of the research indicate that the method of clustering patients based on their characteristics is feasible, and using Neural Network can effectively detect the environment variations. Also it is feasible to pre-programmed the better scheduling rules that can be switched in different environments for reducing patients’ waiting time and maintaining resources utilization at proper level.