透過您的圖書館登入
IP:54.234.124.70
  • 學位論文

資料探勘技術在血液透析之應用

Applications of Data Mining Technology to the Hemodialysis

指導教授 : 龐金宗 葉佳炫
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2005年台灣腎臟醫學會統計,台灣地區於2004年底罹患慢性末期腎臟疾病且必須接受腎臟替代治療的病患人數為38,709人,其中接受血液透析治療的病患占92.3%。末期腎臟疾病的病患通常會經歷各種的合併症,並造成病患的罹病率及死亡率,而腎性骨病變是長期血液透析患者常見的一項合併症。高血磷會導致血液透析病患出現繼發性副甲狀腺機能亢進以及不正常的鈣磷代謝,進而造成腎性骨病變。 本研究的目的在運用資料探勘技術找出長期血液透析病患之特徵,同時病患須具有如下的條件:接受至少六個月以上的血液透析治療,同時合併有血清磷(phosphorus)大於或等於5.5 mg/dl、血清鈣磷乘積(calcium phosphorus product)大於或等於55 mg2/dl2、副甲狀腺素(intact parathyroid hormone)大於或等於300 pg/dl。本研究自三家血液透析中心蒐集550位的病患資料,將其中254名符合上述篩選條件的病患作為研究對象。研究中使用IBM所開發的Intelligent Miner軟體作為資料探勘工具,並運用叢集分析與決策樹分析,以辨認該群研究對象的特性。 資料探勘結果所得到的病患特性依其重要性排列如下:活動能力佳、年齡43.5到64.5歲、原發病為慢性腎絲球腎炎、血清三酸甘油脂正常、長期透析治療年數為7到12年,以及良好的營養狀況。本研究提出以資料探勘的方法,描述出容易併發腎性骨病變的長期血液透析病患之特性。因此,醫護人員較容易審視這群病人,而且醫護人員可以提供病患適當的醫療建議並且擬定策略以預防長期腎性骨病變的發生。

並列摘要


The Taiwan Society of Nephrology 2005 Annual Report showed that in 2004, 38709 people in Taiwan with chronic end stage renal disease (ESRD) required renal replacement therapy. Of these ESRD patients, 92.3% received long-term hemodialysis. Patients with ESRD commonly experience several complications contributing to morbidity and mortality, and renal osteodystrophy is a common complication in the hemodialysis patients. Hyperphosphatemia in patients undergoing hemodialysis usually leads to secondary hyperparathyroidism, abnormal calcium-phosphorus metabolism, and the consequent renal osteodystrophy. The purpose of this study was to use data mining technology to find out the characteristics of patients who had received hemodialysis for at least 6 months in conjunction with serum phosphorus levels≧5.5 mg/dl, serum calcium-phosphorus ion product (Ca × P)≧55 mg2/dl2, or serum intact parathyroid hormone levels≧300 pg/dl. A total of 550 patients were collected from three hemodialysis centers. Among those patients, 254 patients fulfilled aforementioned criteria were enrolled in this study. Software used was the Intelligent Miner developed by IBM, and the characteristics of these patients were identified by clustering and decision tree analyses. The analytic results in order of importance were as followings: good daily activity, aged 43.5 to 64.5, chronic glomerulonephritis as the etiology of ESRD, normal serum triglyceride levels, continuance of hemodialysis for 7 to 12 years, and better nutritional status. This study proposes data mining as a method to delineate the characteristics of long-term hemodialysis patients prone to renal osteodystrophy, and therefore medical staff are able to observe these patients easily. Moreover, medical staff can give them appropriate recommendations and work out strategies to prevent the development of renal osteodystrophy in the long run.

參考文獻


[19]吳長洲,「運用資料探勘技術在產險行銷之個案研究」,私立中原大學,碩士論文,民國九十三年。
[28]紐淑芬,「透析病患之生活品質及其相關因素探討」,台灣腎臟護理學會雜誌,88~101頁,民國九十三年十一月。
[32]黃仁貴,「以叢集分析技術探討病患就診屬性與看診時間之關係」,私立台北醫學大學,碩士論文,民國九十一年。
[3]Chao-Fu Chang, et al.”Comparative Study of Sevelamer Hydrochloride Versus Calcium Carbonate in Hyper- phosphatemic Hemodialysis Patients: An Open-Label, Randomized, Active Controlled, Parallel Study”, Acta Nephrologica, Vol.21, No.1, 2007.
[1]”International comparisons”, USRDS Annual Date Report, pp.218-224,2004.

延伸閱讀