智慧型藥品需求量預測專家系統之建置

The Implementation of an Expert System for the Demand Prediction of Pharmaceutical Drugs

賴順益(Shun-Yi Lai)

指導教授 ： 廖岳祥

亞洲大學/資訊學院/資訊工程學系碩士在職專班/碩士(2011年)

摘要

全民健保制度實施後，面對國內醫療環境的變遷，醫療院所除了提升服務品質，增加病患回診率之外，還要思考如何撙節成本的具體作為。藥品成本在醫院營運成本中，佔有相當重要的地位，如何有效管理藥品庫存，進而降低庫存成本、提高藥品週轉率，儼然為亟需受重視的議題之一。本研究以2006至2009年藥品消耗量資料作為分析數據，並探討目前醫院藥品訂購系統的現況，進而改善現行藥品需求量預測機制。

據過去研究顯示，藥品消耗態樣，不應以齊頭式的方法計算。因此，本研究按醫院的藥品消耗數據，透過5種預測方法及4種檢測預測方法，探討出五種消耗型態的藥品訂購預測模式，並整合支援向量機(Support Vector Machine, SVM)及決策樹，規劃出一套智慧型藥品需求量預測專家系統，提供作為醫院藥品訂購量預測之參考。

關鍵字

藥品消耗態樣 ； 預測模式 ； 專家系統 ； 支援向量機 ； 決策樹

並列摘要

The impact of nationwide health insurance system has significantly changed our domestic health care environment. In addition to improving the quality of medical service and increasing patient return visits, many medical institutions strive to cut costs while maintaining a high quality of service to patients. The cost of the medication inventory is one of the critical factors in total hospital operating costs. An effort to effectively manage the medication inventory, to minimize inventory costs, and to improve the medication flow process has drawn increasing attention to many medical businesses. In this study, we collected four years of data of the consumption of medication, from 2006 through 2009, conducted a statistical data analysis, and evaluated the current ordering system of medication from many hospitals. Thereby, we developed a forecasting methodology to improve the existing mechanism of the medication ordering system.

According to past studies, the consumption patterns of medications are dynamic and unpredictable, thus it is not easy to predict the medication demands in one standard algorithm. In this study, we analyzed four years of medication consumption data from hospitals and used five forecasting methods along with four testing algorithms to develop five forecasting models for hospitals to use in predicting and ordering their medication inventory. In addition, we integrated SVM (Support Vector Machine) with a decision-making tree and developed an expert system that forecasts the demand of medication. It provides hospitals with a helpful tool to make decisions and to order their medication inventory.

並列關鍵字

Medication Consumption Pattern ； Forecasting Model ； Expert System ； Support Vector Machine ； Decision Tree

參考文獻

[8] 陳景堂，統計分析SPSS for Windows入門與應用，儒林圖書，2005。

[13] 何佩芸，醫療供應鏈之協同規劃、預測、補貨模式建構研究，台北科技大學商業自動化與管理研究所碩士論文，2005。

[16] Chih-Chung Chang and Chih-Jen Lin，LIBSVM -- A Library for Support Vector Machines，http://www.csie.ntu.edu.tw/~cjlin/libsvm/。

[1] 袁立德，藥品消耗型態與庫存管理之實證研究－以二所群醫中心為例，國立陽明大學醫務管理研究所碩士論文，1994。

[2] Michael Negnevitsk著，顧力栩、沈晉惠等譯，人工智慧：智慧型系統導論(第二版) ，新文京開發出版社，2007。