近年來,慢性腎臟病(Chronic kidney disease, CKD)的疾病管理政策已成為我國衛生單位相當受重視的公共衛生議題,也是目前世界各國衛生單位努力防治的慢性疾病。依據衛生福利部中央健康保險署2015年的統計,台灣地區慢性腎臟病於全民健康保險醫療費用的總支出佔10大疾病的首位,而血液透析支費用更是高達平均每人一年花掉將近70萬元。因此,藉由建構影響慢性腎臟病患病程進展之預測模式,以提出末期腎臟病發生率的預防措施及建議,進而降低醫療費用的負擔,此乃刻不容緩之重要議題。一般在臨床上都會透過定期追蹤管理來監控病患的生理狀況,掌握可能影響CKD病程惡化的因素並進行相關治療,以延緩病程進展的速度來降低末期腎臟病的發生。然而,在過去評估CKD病程進展的相關研究,其多以傳統統計之方法做為評估與衡量之工具,甚少運用大型資料庫之方法加以分析與衡量。因此,本研究首先透過文獻建構評估慢性腎臟病病程進展之預測模式,之後以2004至2015年台灣南部某醫學中心CKD收案病患為研究對象,運用資料探勘不同分類器之方法找出最適之預測工具,最後運用最適之方法評估CKD病患一年後之病程進展。研究結果發現,AdaBoost+決策樹為最適之預測工具,並歸納出早期發現病情與接受治療,且利用個案管理加強病情的控制與衛教的宣導,是可以降低病情惡化的結果。本研究期望所建構之模式可提供臨床與實務面一個重要的參考依據。
In recent years, the disease management policies of chronic kidney disease (CKD) has become an important public health issue for Taiwanese medical section and it was also the World Health Organization effort to combat chronic diseases. According to the reports in 2015 from National Health Insurance Administration Ministry of Health and Welfare, Taiwan, the expenditure of chronic kidney disease is the first on top 10’s disease and the average of expense of dialysis for each person was spend NT 700,000 dollars/per year. Therefore, establish and propose an evaluation model to reduce incidence of CKD for end-stage renal disease is an important issue. The tradition evaluation model was focus on statistical theory to evaluation CKD risk, very few study was used data mining to build and evaluate the CKD risk. In this study, first we present an evaluation model through the literature review. Next, the study was aimed to 2004 to 2015 the patients of CKD at the medical center in southern Taiwan to identify proposal model and the optimum method. Finally, using proposal model and the optimum method to evaluate impact factor for retarding progression of CKD. The results show that AdaBoost + tree method is able to produce good outcome. Besides, if the CKD patients can provide early treatment and nursing case management plan, the outcome would be reduce the growth of the disease. Importantly, the proposed model can assist the medical sector to assess the organizational performance of hospitals, making it highly applicable for academia and clinic medicine purposes.