運用資料探勘技術預測洗腎患者之透析廓清率

美國腎臟疾病資料庫(United States Renal Data System, USRDS)於2011年發表的年報中指出:台灣洗腎發生率347人/百萬人列居世界第二名，盛行率約每百萬人口2447人，高居世界排名第一位(USRDS,2010)，根據台灣腎臟醫學會統計2012年統計末期腎病變接受透析治療的總人數全台有六萬三千六百四十二人數，其中五萬七千二百八十四人是接受血液透析佔全國90%，六千三百五十八人接受腹膜透析治療佔全國10%。顯示血液透析為末期腎病患者主要選擇之替代療法，美國國家腎臟基金會(National Kideny Foundation, NKF)的腎臟病療效與品質研發組織(Kideny Disease Outcome Quality ,KDOQI)所訂的透析照護指引中建議接受血液透析患者的廓清率應該維持在1.2以上才能降低合併症及住院率進而減少死亡率發生。本研究運用模式樹(M5)、類神經網路(ANN)、支援向量機(SVM)等技術，預測透析治療廓清率的影響因子，為了提昇預測準確度，進一步將Bagging效能提昇技術加入實驗中，最後整體評估結果顯示，三種預測模式中M5模式樹加上Bagging效能提昇所建構之分類器預測效能最佳。本研究所建構的預測模式可以提供臨床照護血液透析患者調整透析處方的參考，提昇透析患者生活品質。

關鍵字

模式樹；類神經網路；洗腎

並列摘要

United States Renal Disease Database (United States Renal Data System, USRDS) report published in 2011 stated: Taiwan dialysis incidence of 347 people / million people out ranks second in the world, the prevalence of approximately 2447 people per million population , the world's highest ranked one (USRDS, 2010), according to the Taiwan Society of Nephrology Statistics 2012 Statistics of ESRD patients on dialysis, the total number of Taiwan have sixty-three thousand six hundred forty-two, among them fifty-seven thousand 284 hemodialysis is 90% of the country's total, and six thousand 358 receiving peritoneal dialysis accounted for 10%. Show hemodialysis ESRD patients as the main choice of alternative therapies, the National Kidney Foundation (NKF) and Kideny Disease Outcome Quality( KDOQI) guidelines laid down in the proposal received dialysis care clearance of hemodialysis patients should be maintained at 1.2 or more to reduce complications and hospitalization rates and thus reduce mortality occurred. This research uses the model tree (M5), artificial neural network (ANN), Support Vector Machine (SVM) techniques to predict the impact of dialysis clearance factor, in order to improve forecast accuracy, and further enhance the effectiveness of Bagging technology into the experiment Finally, the overall results of the assessment showed that three kinds of prediction modes M5 model tree constructed by adding Bagging enhance the performance of the classifier prediction performance optimization. Prediction model constructed in this study can provide clinical care dialysis prescription in hemodialysis patients adjust reference to enhance the quality of life in dialysis patients.

並列關鍵字

model tree ； artificial neural network ； dialysis

參考文獻

胡連順、尤俊成、洪振傑、徐永堂(民 100)。淺談人工腎臟與透析膜構造[1]:人工腎臟發展史。腎臟與透析。23(1)。65-70。

Hsu, C. W., Chang, C. C., & Lin, C. J. (2009). A practical guide to support vector classification, 2003. Paper available at http://www. csie. ntu. edu. tw/ cjlin/papers/guide/guide. pdf.

郭志隆、張芳綾(譯)(民 97)。Introduction to Business Data Mining。(原作者David Olson,Yong Shi)。台北市：美商麥格羅．希爾國際股份有限公司台灣分公司。

馮英鳳、邱小鳳(民 100)。洗乾淨了沒-談適量透析概念。台灣腎臟護理學會雜誌。11(1)。24-26。

Beige, J., Sharma, A. M., Distler, A., Offermann, G., & Preuschof, L. (1999). Monitoring dialysis efficacy by comparing delivered and predicted Kt/V. Nephrology Dialysis Transplantation, 14(3), 683-687.

國際替代計量

運用資料探勘技術預測洗腎患者之透析廓清率

未授權

主題瀏覽