以馬可夫鏈模型預測醫院建築非定期修繕成本之研究

Abstract

醫院建築規模與設備之複雜度與精密度皆高於一般建築物，一旦疏於修繕維護，除支付硬體之高額金錢損失，甚至會造成人命的傷亡。近期台大醫院意外災害之發生，凸顯醫院建築於修繕維護的重要，而醫療設備長時間提供服務與實驗研究使用，其硬體之良莠對於完善醫療佔有不可或缺的重要地位。近年景氣不佳使國家補助預算逐年劇減；短絀的預算且缺乏預測維護成本模型，各方問題已導致醫院硬體成本管控上之困難。而醫院建築針對修繕維護相關的研究文獻甚少，且多以訪談、問卷等方式探討，尚無以實際資料進行預測。本研究以台灣大學醫學院附設醫院東址院區為研究對象，藉由醫院修繕維護之長期資料分析，針對非定期修繕資料作出預測，參考其準確性用於醫院的維護與管理。 馬可夫鏈鎖過程（Markov Process）為一種可用於預測之方法，是以資料本身來決定推估其變動之形態，並由分析過去一段時期中之變動過程及其結構，推測未來發生的可能性。本研究取得近十年共5,028筆實際修繕維護數據，進行專家訪談；並為求增加預測準確性，將實際修繕維護數據依不同使用年齡之屬性以及金額級距加以區分，以馬可夫鏈（Markov Chains）進行預測，並將結果作分析比較，建構醫院建築之非定期修繕成本模型。 本研究之預測模型結果提供醫院工務室主管一簡單而具有依據之方式，以進行未來年度實際成本之預測與估算，使醫院管理者能有效掌握資金分配，更作為未來申請預算之合理依據，並藉由預測結果凸顯目前預算編列制度不合理之處。

The architectural functions of a hospital building are more complex than commercial or residential building. Once the hospital building maintenance is neglected, not only will there be a high cost of repair but also a loss of human lives. A recent accident at National Taiwan University Hospital (NTUH) has revealed the significance of hospital building maintenance. For providing long-term medical services, the hospital equipments play an irreplaceable role in quality assurance of medical care. As the economy suffers in recent years, the government’s fund for hospital preservation also declines. Consequently, there is an increase in difficulty to manage a hospital. A few publications have initiated interviews and surveys, but lacks of historical records for prediction. Using the East Site of Taiwan National University as the case study, this research analyzes the current hospital building maintenance record and predicts non-periodical maintenance cost for future. Markov Process is a prediction method that utilizes present information to describe future condition. The future condition is assumed by reviewing the changes of the past years. With detail review of 5,028 records of hospital building maintenance over the last ten years, this study proposes a cost prediction model for non-periodical maintenance by using Markov Chain as the methodology. Variables including frequency and cost are also taken into consideration to enhance accuracy. This hospital building maintenance prediction model could be a reference for hospital management to budget more effectively for the coming year. It can also check the rationality of present budget mechanism.