Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/98252
標題: 以監督式學習及資料重整技術於預測慢性腎臟病之研究
A Study on Chronic Kidney Disease Prediction Using Supervised Learning Methods and Data Cleaning Skills
作者: 黃吉立
Chi-Li Huang
關鍵字: 監督式學習;資料重整技術;慢性腎臟病;Supervised Learning Methods;Data Cleaning Skills;Chronic Kidney Disease
摘要: 
台灣腎臟醫學會發行的2017年台灣腎病年報內容提到,台灣末期腎臟病(End Stage Renal Disease,ESRD)的發生率仍然為世界第一,遠高於歐、美洲和日本等國家;洗腎的盛行率及發生率,亦是世界第一。以衛生福利部中央健康保險署醫療費用統計,急、慢性腎臟病花費四百六十億元,比例佔全民健保總支出的7.31%;而全國洗腎總人口數則是增加至八萬五千人,也都創下歷史的新高記錄。在現今台灣社會開始正式歩入「高齡化社會」的環境下,慢性腎臟病對國家及人民的影響是值得嚴肅去正視的重要議題。
本論文研究配合資料重整技術模擬大量醫院病患的不平衡數據,以遞減特徵屬性數量來進行以監督式學習分類演算法的資料探勘並結合醫療資料建置預測疾病的模型應用,嘗試建立一快速可靠的模型,來預測慢性腎臟病,研究結果顯示在四種演算法模型中預測結果準確率都可高達99%,未來有機會以實際檢驗數據來預測,能夠真正協助問診或臨床醫師,正確的診斷分類,能夠及早發現及早治療,減少慢性腎臟病患者,避免政府健保預算超支。

In 2017 Taiwan Nephrology Annual Report issued by the Taiwan Society of Nephrology, the incidence of end stage Renal Disease(ESRD) in Taiwan is still the highest in the world, much higher than in countries such as Europe, America and Japan. Prevalence and incidence are also the highest in the world. According to the medical expenses of the Central Health Insurance Department of the Ministry of Health and Welfare, acute and chronic kidney diseases cost 460 billion NT dollars, accounting for 7.31% of the total health insurance expenditure. The total number of dialysis patients nationwide has increased to 85 thousand, which is also a record high in history. In today's Taiwanese society, which has officially entered the 'Aging society ', the impact of chronic kidney disease on the country and the people is an important issue that deserves serious consideration.
This thesis studies the data cleaning skills to simulate the imbalance data of a large number of hospital patients, and reduces the number of feature attributes for and combines medical data to build supervised learning algorithm for predicting disease, and attempts to establish a fast and reliable model to predict Chronic Kidney Disease, the results show that the accuracy of the prediction results of the four algorithms models can be as high as 99%. In the future, there is an opportunity to predict the actual examine data, and it can really help the doctor or clinician to make correct and reliable diagnosis. early detection and early treatment, reducing chronic kidney disease patients, avoiding government health insurance budget overruns.
URI: http://hdl.handle.net/11455/98252
Rights: 同意授權瀏覽/列印電子全文服務,2022-01-31起公開。
Appears in Collections:資訊管理學系

Files in This Item:
File SizeFormat Existing users please Login
nchu-108-5104029012-1.pdf1.32 MBAdobe PDFThis file is only available in the university internal network    Request a copy
Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.