結合FCA與CBR方法應用於第二型糖尿病併發症診斷模式建立

Constructing a Model of type II Diabetes Complications Diagnosis Using FCA and CBR Approaches

柯宜君(Yi-Chun)

指導教授 ： 林榮禾

國立臺北科技大學/管理學院/工業工程與管理研究所/碩士(2011年)

摘要

隨著社會經濟的繁榮及國民生活型態的改變，目前全球己有一億三千五百萬的糖尿病人口，學者預估到西元2025年全球糖尿病人口成長率將高達35 %。在台灣地區糖尿病是高居於國人十大死因排行榜的第四名，而糖尿病的危害主要來自併發症，長期血糖控制不當，容易引起許多併發症，在國人十大死因當中，有半數死因與糖尿病併發症有關。故其併發症研究是目前重要議題之ㄧ。
本研究目的為建立兩階段全面性第二型糖尿病患者併發症診斷模式，利用病歷結構擷取部份面向架構，來分析「糖尿病患者症狀出現」所可能產生的倂發症，將典型的症狀利用正規概念分析法(FCA)處理，將症狀與併發症產生子概念與父概念的關係，透過這個關係建構出併發症模式，並結合決策樹輔助診斷分析，作為第一階段糖尿病患者是否罹患併發症的判別模式；第二階段再結合案例式推理(CBR)，推導得出相似度高的病患案例資料，案例內容會有特徵描述和解決方案二個部份，從查詢後獲得的案例中得知解決方案。
研究結果顯示，決策樹C5.0併發症判別模式其準確率達97.92%，而且CBR除推導出相似案例外，亦能利用改編案例規則協助患者明瞭其他併發症的可能性，以達成醫師在併發症診斷推論及治療之實務有效應用；透過這個架構引導，藉以避免糖尿病患者併發症發生的可能性，並能及早發現，及早治療，以減少醫療上的損失，提升醫療品質。

關鍵字

糖尿病併發症診斷 ； 正規概念分析法 ； 決策樹 ； 案例式推理

並列摘要

As society and economy growing and the citizens’ lifestyle changing, there are one hundred thirty-five million people with diabetes around the world. Scholars predicted that the world population growth rate of diabetes will be up to 35% around 2025. In Taiwan, diabetes is the fourth leading cause of death. It caused lots of complications. The damage mainly from complications of diabetes is inappropriately long-term glycemic control which easily leads to many complications. Among the national top ten causes of death, half of the cause of death is related to diabetic complications. Therefore, the issue of diabetic complications is one of important studies currently.
The purpose of this study aims to establish a comprehensive two-stage diagnosis of complications in patients with type II diabetes, using medical records for the structure extraction part of the framework to analyze the Type II diabetes symptoms that may develop complications. Typical symptoms of the concept of using formal concept analysis (FCA) treated under the existing medical structures. Most of the content of medical records used text descriptions that belong to semi-structured documents. The study cited International Classification of Primary Care (ICPC), International Classification of Diseases (ICD) of medical terminology and the formal concepts provided a method for structuring and displaying the relationships between the symptoms and complications. The complications of this relationship through the construction of models that combined with the decision tree analysis, as the first phase of the diagnostic mode.
Combined with the second phase of case based reasoning (CBR), it will derive highly similar scores cases and content will show feature case description and solution of two parts. The results showed that, by means of C5.0 decision tree determine complications, the accuracy rate of its model is 97.92%. And similar cases are derived by CBR, the adaptation case using the rules also helps patients understand the possibility of other complications which they may develop. With the view of this framework, the purpose is to help physicians make practical use of the treatment of complications. It may avoid the possibility of complications with diabetes and reduce the loss of medical care even improve the quality of its.

並列關鍵字

Type II Diabetes diagnosis ； FCA ； decision tree ； CBR

參考文獻

[22] 陳憶萱，智慧型醫療診斷模式建構於肝病診斷應用，碩士論文，台北科技大學商業自動化與管理研究所，2006。

[6] 行政院衛生署國民健康局，台灣糖尿病防治推動現況與展望，台北：行政院衛生署國民健康局，2006。

[10] 行政院衛生署國民健康局，台灣糖尿病血糖與併發症研究，台北: 行政院衛生署國民健康局，2002。

[5] 戴東原，糖尿病在台灣的現況•中華民國內分泌暨糖尿病學會會刊，第二十一期，2008，第8-9頁。

[18] 廖麗琴、歐香縫、劉波兒、王潤清、程培鈺，「良性陣發性姿勢性眩暈病患生活品質初探」，志為護理雜誌，第七卷，第二期，2008，第70-78頁。