決策支援系統－以血液透析管路存活率分析為例

Decision Support System－A Case Study of Hemodialysis Tubing Survival

曾秀蓮(Siou-Lian Zeng)

指導教授 ： 黃光宇

嶺東科技大學/管理學院/高階主管企管碩士在職專班/碩士(2015年)

https://doi.org/10.6823/LTU.2015.00001

摘要

本研究提出一種混合式的資料挖掘方法，稱為HD-index方法，並結合多元線性迴歸分析來印證並用於實際醫學應用上。HD-index方法中包括：Fuzzy C-Means (FCM)分群方法、指標函數（Index function）、粗集合理論（Rough Set Theory，RST）、MH-index叢集效度指標函數。將連續值離散化後，再分群依前面提到指標去判斷連續值是屬於何群組；透過資料挖掘，發現資料中隱含的趨勢或者模式，將原始資料集先過濾雜訊後成為精準資料集，採用多元線性迴歸，找出影響血液透析管路存活相關因素，其結果做為決策考量依據。而透析管路重建，相對增加健保醫療支出及直接照護人力需求，尤其在護理人力嚴重短缺情況下，為健康照護體系帶來嚴重的困擾。因此，本研究目的為探討台北市某四家洗腎中心，影響血液透析管路功能重要相關因素分析與預測，研究個案193位，長期使用動靜脈瘻管血液透析之病患。採回溯性研究，運用病歷回顧方式，以編碼匿名收集病患基本資料，包括：伴隨的疾病、生活習慣、透析照護及人口學特性；採用SPSS 21版多元線性迴歸中之逐步迴歸分析法，進行研究假設之驗證。研究結果：原始及精準的血液透析資料集：所有模式p-value <0.05為顯著相關，依變數血液透析管路存活率和患有糖尿病、膽固醇平均值呈現負向影響；目前使用自體瘻管型態呈現正向影響。因此，可將此列為高危險群做為臨床照護依據，以降低血液透析管路重建所延伸醫療支出與護理照護人力之問題。

關鍵字

血液透析 ； 血管通路 ； HD- index函數 ； 決策支援系統

並列摘要

This study proposes a hybrid data mining method, called HD-index method, and multiple linear regression analysis to prove and to practical medical applications. HD-index method includes: Fuzzy C-Means (FCM) clustering method, index function (Index function), rough set theory (Rough Set Theory, RST), MH-index cluster validity index function. The continuous values discrete, then grouped according to the aforementioned indicators to determine what belongs to the group of continuous value; after digging through the data and found that the information implicit in the trend or pattern, the original data set to filter noise become accurate information sets, using multiple linear regression to find Hemodialysis pipeline survival factors, consider the results as a basis for decision-making. The dialysis tubing reconstruction, the relative increase in health care expenditure and direct medical care manpower needs, especially in the nursing workforce shortages, the health care system has serious problems. Therefore, the purpose of this study was to explore a four dialysis centers in Taipei, Hemodialysis pipeline functionally important factors analysis and forecasting, case study 193, patients receiving long-term hemodialysis of venous fistula move. Mining retrospective study using chart review (chart review) way to encode anonymously collect basic patient information, including: concomitant diseases, habits, and demographic characteristics of the dialysis care; using SPSS version 21, stepwise multiple linear regression of regression analysis to verify the hypothesis. The result: the original and accurate data collection hemodialysis: All modes p-value <0.05 was significantly related, according to the variables dialysis tubing survival and suffering from diabetes, cholesterol mean presenting negative effect; the current use of autologous fistula type state presents a positive influence. Therefore, as this high-risk group as the basis for clinical care to reduce health care spending hemodialysis pipeline reconstruction and nursing care of the problem of human extension.

並列關鍵字

hemodialysis ； vascular access ； HD- index function ； decision support system

參考文獻

【31】Shah, R., Bhatt, U. Y., Cleef, S. V.,Farley, M., Davis, A., Swope, M., &Agarwal, A. K. (2011). Vascular accessthrombosis and interventions in patientmissing hemodialysis sessions. Clinical Nephrology. 76(6), 435-439. doi.org/10.5414/CN107086.

【33】Salmela, B., Hartman, J., Peltonen, S.,Albäck, A., & Lassila, R. (2013).Thrombophilia and arteriovenous fi stulasurvival in ESRD. Clinical journal ofthe American Society of Nephrology.[Epub ahead of print]. doi:10.2215/CJN.03860412.

doi:10.1038/ki.2008.318; published online 30 July 2008

【43】Gagliardi, G. M., Mancuso, D., Falbo, E.,Mollica, F., Mollica, A., Barcellona, E.,Bonofiglio, R. (2012). Anthropometric parameters of nutritional assessment as predictive factors of arteriovenous fistula malfunction in patients undergoing hemodialysis. The Journal of Vascular Access, 13 (4), 475-481.doi: 10.5301/jva.5000096

【47】Monroy-Cuadros M, Yilmaz S, Salazar-Bañuelos A, Doig C.(2010)Risk factors associated with patency loss of hemodialysis vascular access within 6 months. Clin J Am Soc Nephrol. 5(10) ,1787-1792. doi: 10.2215/CJN.09441209. Epub 2010 Jun 24. Oct;5(10):1787-1792.