| 摘要: | 在今日醫療機構競爭日趨激烈，以及政策與制度多元化的醫療環境中，醫院管理者在面對複雜的決策問題時，除了參考專業知識與過去的決策經驗外，更應熟悉如何應用資訊技術以輔助問題解決及決策制定。 過去決策支援系統已有應用在醫療或醫院管理的例子，雖然系統建置方法上有不同的差異性存在，最常見的方法仍是為特定決策目標設計與開發，但是如果抽離該特定決策模式，則顯得適用性不夠。因此，為了克服過去決策支援系統無法任意修改決策目標的性質，我們提出一個構想，意圖建立一個非特定用途之決策支援系統，再由管理者制定其決策知識，賦予該系統的決策任務。所以在研究中，我們同時也提出決策因子與決策知識可以多重遞迴的觀念，以輔助我們做到決策目標多元化之目的。 因此在本篇論文所設計的決策支援系統，可以讓使用者選擇其偏好的決策目標，篩選符合決策目標的資料項目，不但可以藉此決策知識解決決策上之問題，甚至可以讓使用者建立屬於自己的決策知識或是決策模式。而在決策支援系統架構部分，我們允許決策目標由多個決策因子所構成，更可以支援決策目標由多重遞迴關係式的決策因子所構成。 在系統驗證過程中，我們採用醫療方面糖尿病的篩選與醫管方面的門診相關資料統計。從驗證結果得知我們的系統確實可以正確無誤的篩選出符合決策目標的資料項目；除此之外，從多達約400萬筆檢驗資料中篩選糖尿病的例子，我們也發現到一個事實，當資料量越龐大時，雖然總運算時間也隨之增加，但是每一筆資料項目所花費的單位時間則呈現遞減的現象，說明了系統在資料量越龐大時，所需的運算時間相較於資料量小時更具效率。; In this medical conditions, due to the competition among the medical institutions are getting harder and the policy are diversified. The hospital managers, when face the complicate decision problem, shall refer to not only the professional knowledge and experience but also the aid of IT to assist us in decision making and problem solving. In the past, the decision support system had been developed in the medical area or hospital management. Although the ways of establishing are different, the most common way is to design and implement for particular decisive purpose which suitability is limited if we didn’t apply on particular decisive purpose. For overcoming the past decision support system, which can’t change decisive objective arbitrarily, we propose to establish a decision support system for un-specific purpose, then manager make decisive knowledge and determine decisive purpose. Therefore in this paper, we also bring up an idea, that the decisive factors and decisive objectives could be recursive, to assist us in attaining the goal of decisive purpose diversified. The decision support system we recommend in this paper could let user chose favored decisive purpose and sift the conformed data, we can solve problems by this way and let user set up themselves decisive purposes. In the structure of decision support system, we allow the decisive purpose, which compose of plenty and recursive decisive factors. In the verified aspect, we adopt the diabetes example of medical area and the outpatient services statistics of hospital management. From the result, we know our system could sift the conformed data correctly; additionally, we also confirm that if amounts of data are more and more and the total time of processing time increase as well, but the time of an unit of data spending appear in decreasing. This thing explain the system spend time of processing on larger amount of data less then smaller amount of data. |
