摘要
由於現代人工作的忙碌，生活飲食型態改變與缺乏適度的運動，以至於糖尿病盛行率快速的增加。糖尿病是一種無法治癒的慢性病，但是透過適當照護可以有效的控制病情與避免併發症的發生。糖尿病照護計畫需要根據個人的情況來設計包含藥品、生理檢查、飲食、活動、心理輔導等不同專業領域內容的照護計畫，如果是以人工方式設計照護計畫，常常會受限於評估者的經驗、學識等方面的差異，而無法得到符合個人需要的照護建議；而且，在現今健保資源緊縮的環境中，也希望夠降低昂貴的照護人力資源消耗，以因應高齡化社會所增加對照護人力的需求。 本研究的目的主要是以人工智慧的技術來協助糖尿病照護計畫的擬定，首先，我們以案例式推理技術根據糖尿病病患的情況進行個人化照護計畫的搜尋。透過案例式推理，可以應用過去解決問題的經驗來處理目前的問題；而且經由案例修改來更加符合目前的狀況後，可以將此案例存入知識庫中，不斷的擴充知識庫，使系統往後的照護建議更加符合使用者所需。再來，本研究進一步以本體論為基礎，將糖尿病照護所需要的各相關知識，直接且結構性的建置出糖尿病照護本體論。當案例式推理模組沒有適合的案例可以使用時，則由糖尿病照護本體論模組推理出基礎的照護建議，再經由照護專家調整之後變成新的案例儲存在案例資料庫，並將此案例回饋給糖尿病照護本體論模組。利用本體論的好處是可以發揮知識的共享性與可再利用性。研究結果顯示，本系統可以提供糖尿病個人化照護的決策建議，降低消耗醫療資源，維持照護建議的品質。未來還可以擴充到其他慢性病照護，終能達成一個可以提供各種疾病照護建議的決策支援系統，並且能提供新進醫護人員教育訓練與照護建議範本。
並列摘要
In recent years, because of changing in lifestyle and diet, and lacking of moderate exercise, diabetes prevalence increases rapidly. Diabetes is an incurable chronic disease, but through proper care, patients can control the disease and prevent complications. A proper diabetes care plan requires different fields of professionals together to make up the plan according to patient’s personal need. So if it is done manually, it would restrict to the experience and knowledge of these professionals, and consume lots of these expensive medical resources as well. So how a timely and effective diabetes care plan can be made and to meet the personalized needs is a very important research issue. The main purpose of this study is to use artificial intelligence technique to generate personalized diabetes care plan. To do that, we use case-based reasoning to search matched diabetes care plan according to personal condition. This plan can be used directly or modify it to better fit current problem. The modified case is saved in our case database for next time usage. The expansion of case database making the system can propose care plan more in line with user requirement later. When there is no matched case, embedded in our system is a diabetes care ontology which can be consulted. The benefit of using ontology is knowledge of diabetes care can be shared and reused easily. The ontology generated a basic care plan to care professionals for adjustment. Again, this adjusted plan becomes the new case and our system will deposit it in case database. The result of our research shows our system can provide personalized diabetes care plan efficiently. In the future, this system can be expanded to other diseases care and eventually becomes a total diseases care decision support system