透過您的圖書館登入
IP:18.188.20.56
  • 學位論文

結合BPN、AHP與CBR方法建構於智慧型肝病醫療診斷模式

Constructing an Intelligent Liver Diagnosis Model Using BPN, AHP, and CBR Approaches

指導教授 : 林榮禾
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


根據衛生署2000年至2006年統計台灣十大死因資料顯示,癌症位居十大死因之首,其中肝癌更是台灣第一大癌症病因。而慢性肝病及肝硬化為台灣十大死因的第七位。肝病初期症狀不明顯,即使是肝硬化或肝癌等嚴重的肝病,恐怕也要等到病情相當嚴重時才會出現症狀。而過去肝病診斷研究大多以單一肝病類型為主,如慢性活動性肝炎與肝癌,尚無針對全面性肝病類型之診斷模式建立。因此為達到更準確與全面性的肝病醫療診斷及提昇醫療品質與降低診斷成本,本研究之目的為建立智慧型肝病醫療診斷模式 (ILDM),並收集病患之生化檢驗資料及其生活習慣健康調查資料,運用倒傳遞類神經網路 (BPN) 建構肝病判別模式,來判定是否罹患肝病,再結合層級分析法 (AHP) 及案例式推理 (CBR) 找出指標屬性權重值,來診斷病患罹患肝病類型與其治療方式。研究結果顯示,BPN肝病判別模式其分類平均準確率達98.17% (訓練準確率100%,測試準確率93.94 %),且CBR的肝病類型診斷準確率達100%。另外,醫師專家以AHP法制定之權重值亦能增加CBR診斷之可信度,並有效提出肝病之藥品採用及治療方法,且醫師可依病患生活習慣改善建議,以輔助肝病病患之治療。CBR除分析罹患肝病類型之可能性外,亦能避免BPN造成誤判病患結果,以達成醫師在肝病診斷推論及治療之實務有效應用。

並列摘要


According to the statistics of Department of Health (DOH) from 2000 to 2006, the cancer is the first disease of the top ten fatal diseases. Among them, the liver cancer is the first cause of disease of the cancer in Taiwan, and the chronic liver diseases and cirrhosis are the seventh of the top ten fatal diseases in Taiwan. The symptoms of liver diseases are not obvious in the initial stage. Even serious liver diseases, such as cirrhosis or the liver cancer, the symptoms will not appear until the condition is quite serious. Most studies of liver diagnosis in the past relied mainly on single type of liver disease, such as the chronic activity hepatitis and liver cancer. There is no liver diagnosis model for all types of liver diseases yet. In order to reach more accurate and complete liver diagnosis, promote medical quality, and reduce the cost of diagnosis, the purpose of this paper is to construct an Intelligent Liver Diagnosis Model (ILDM). Using Back-Propagation Network (BPN), patient’s laboratory data and health survey are collected to construct a liver disease prediction model which is applied to diagnose whether suffer from a liver disease. Then, Analytic Hierarchy Process (AHP) and Case-Based Reasoning (CBR) are combined to find out the weight values of attributes which is used to diagnose the types of liver diseases and to recommend the appropriate prognosis and treatment. The research results indicate that the average accuracy rate of BPN prediction is 98.17% (training set accuracy rate is 100%, testing set accuracy rate is 93.94%) and CBR diagnostic accuracy rate is 100%. Since the professional physicians’ experiences are provided to initialize the weight values of AHP which can improve the credibility of CBR diagnostic, and it recommends the medicines and treatment for the liver diseases more effectively. The physicians can offer suggestions according to patients’ living styles to assist patients’ treatment. Besides of CBR diagnose on probability of liver diseases, it also can reduce diagnostic errors of BPN prediction in order to prove physicians’ diagnose and apply for effective treatment.

並列關鍵字

Liver diagnosis BPN CBR AHP SF-36

參考文獻


[9] 石曜堂、洪永泰、張新儀、劉仁沛、林惠生、張明正、張鳳琴、熊昭、吳聖良,「國民健康訪問調查之調查設計、內容、執行方式與樣本人口特性」,台灣公共衛生雜誌,第二十二卷,第六期,2003,第419-430頁。
[10] 李蘭、黃美維、陸玓玲、潘憐燕、李隆安、鄧肖琳,「台灣地區成人的健康行為探討:分佈情形、因素結構和相關因素」,中華公共衛生雜誌,第十四卷,第四期,1995,第358-368頁。
[12] 胡月娟,「社區民眾健康意識與健康行為初探」,護理研究,第一卷,第四期,1993,第309-315頁。
[14] 劉博仁、李采娟,「過敏性鼻炎患者接受免疫療法之生活品質」,台灣醫學,第九卷,第四期,2005,第455-466頁。
[15] 蔡碩倉,「運用類神經網路建構台灣地區農會信用部金融預警系統」,農業經濟半年刊,第六十八期,第117-156頁。

被引用紀錄


彭晟輔(2012)。環保點膠製程研發與創新參數設計〔碩士論文,國立臺北科技大學〕。華藝線上圖書館。https://doi.org/10.6841/NTUT.2012.00585
留啟祐(2008)。整合資料探勘方法應用於肝病輔助診斷〔碩士論文,國立臺北科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0006-1108200821124700
趙國閔(2011)。應用案例式推理於人工生殖技術懷孕率預測之研究〔碩士論文,國立臺北科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0006-1307201118373600

延伸閱讀