資訊管理學報
邱瑞科;王祥安;吳建樺;張彥群;
頁: 255-281
日期: 2018/07
摘要: 根據行政院衛生福利部國民健康署所公布的民國105年國人主要死因之統計資料顯示，腎炎、腎病症候群及腎病變位居十大主要死因之第九位。又根據2015年美國腎臟登錄系統（the United States Renal Data System；USRDS）年度報告，以洗腎作為指標，台灣末期腎臟病的發生率為世界第一名，而腎臟疾病一般情況下是很難意識到的，許多腎臟疾病之病友發現時，已是末期腎衰竭需要洗腎的情形，因此定期健康檢查及保持良好的飲食及生活習慣是非常重要的，以做到根本之預防，及早發現和及早的治療。本研究模糊專家系統及系統發展工具jFuzzyLogic結合雲端運算服務來建立一套慢性腎臟病檢測及惡化風險評估之雲端資訊服務系統。首先藉由文獻探討及彙整醫師專家的專業知識，以釐清慢性腎臟病之檢測方法及危險因子。之後透過人工智慧之兩種技術包括類神經網路與支援向量機來驗證危險因子之準確度。最後，透過腎絲球過濾率來檢測系統使用者是否有無罹患慢性腎臟病，並進一步運用模糊專家系統推估惡化之可能風險。希冀藉此能發展一套可用於線上檢測及風險評估系統，提供社會大眾作為自我檢測之參考依據與了解患者病情資訊。在雲端平台部署上，本研究最後並將所發展之系統部屬Google公司所提供雲端運算服務平台，以提供使用者進行線上自主性檢測及風險評估。期望能透過雲端運算強大的運算與動態調整基礎建設資源的能力，使得本系統具有更高的彈性、效能、容錯能力與擴充性，來提供更具效能之網際網路系統線上服務。
關鍵字: 慢性腎臟病;人工智慧;模糊專家系統;風險評估;雲端運算服務;
Abstract: Purpose - By the annual mortality statistics announced by the Department of Health Welfare in year 105, nephritis, nephrotic syndrome and nephrosis are ranked among the top ten leading causes of civilian death. Again, according to the 2017 Annual Data Report of United States Renal Data System, it indicated that the mortality rate in end stage renal disease (ESRD) in Taiwan was ranked in the first place in the world. In general, it is difficult to be detected because the symptoms of chronic kidney disease (CKD) are not very apparent. Once ESRD is detected, it may require to rely on the dialysis to maintain alive. Design/methodology/approach - Firstly, through the literature review and interview from experts, the methods of detection and risk factors of chronic kidney disease are determined. Secondly, leverage the factors identified, two artificial intelligent techniques including the back-propagation network (BPN) and support vector machines (SVM) are built and employed to verify the accuracy of these factors of affecting CKD. Lastly, Glomerular Filtration Rate (GFR) is employed to detect whether a system user has CKD followed by the fuzzy expert system is employed to assess the deterioration of probable risk. Findings - Through the aid of this service system, the publics may take advantage of this system over the Internet to conduct an online self-assessment for the risk of suffering CKD. In the meanwhile, the system is deployed on the Google cloud platform to provide a higher performance system online service. Research limitations/implications - Because the volume of patient cases employed in this research are not large enough, the reliability of research outcome may be questioned by the specialty physicians in medical practice even the predication model is thoroughly built and tested. If the source of patient cases can be obtained and extracted from National Health Insurance Research Database (NHIRD), the reliability of outcome would be enhanced to be trusted. Practical implications - The verification and validation of the this research regardless the methods used, model built, and outcomes generated should be furtherly done in order to put the result in practice in medicine domain. In additions, the powerful computing and dynamical adjustment of cloud infrastructure needs also to be cared to make the service system gaining higher flexibility, better performance, greater fault-tolerance and scalability to improve the performance of system online service over the Internet. Originality/value - The early diagnosis and treatment of CKD is extremely important. The study of this paper builds a cloud information service system based on the techniques of cloud computing, fuzzy expert system, and system development tool, jFuzzyLogic The participation of the kidney physician in this research provides the valuable and expert knowledge to build the original and unique predicting model for this research. If the research outcome and methodologies can be furtherly verified and validated, the value of academics and practices of this research can be also expected.
Keywords: chronic kidney disease;artificial intelligence;fuzzy expert system;risk assessment;cloud computing service;
瀏覽次數: 17798 下載次數: 183
引用 導入Endnote
頁: 255-281
日期: 2018/07
摘要: 根據行政院衛生福利部國民健康署所公布的民國105年國人主要死因之統計資料顯示，腎炎、腎病症候群及腎病變位居十大主要死因之第九位。又根據2015年美國腎臟登錄系統（the United States Renal Data System；USRDS）年度報告，以洗腎作為指標，台灣末期腎臟病的發生率為世界第一名，而腎臟疾病一般情況下是很難意識到的，許多腎臟疾病之病友發現時，已是末期腎衰竭需要洗腎的情形，因此定期健康檢查及保持良好的飲食及生活習慣是非常重要的，以做到根本之預防，及早發現和及早的治療。本研究模糊專家系統及系統發展工具jFuzzyLogic結合雲端運算服務來建立一套慢性腎臟病檢測及惡化風險評估之雲端資訊服務系統。首先藉由文獻探討及彙整醫師專家的專業知識，以釐清慢性腎臟病之檢測方法及危險因子。之後透過人工智慧之兩種技術包括類神經網路與支援向量機來驗證危險因子之準確度。最後，透過腎絲球過濾率來檢測系統使用者是否有無罹患慢性腎臟病，並進一步運用模糊專家系統推估惡化之可能風險。希冀藉此能發展一套可用於線上檢測及風險評估系統，提供社會大眾作為自我檢測之參考依據與了解患者病情資訊。在雲端平台部署上，本研究最後並將所發展之系統部屬Google公司所提供雲端運算服務平台，以提供使用者進行線上自主性檢測及風險評估。期望能透過雲端運算強大的運算與動態調整基礎建設資源的能力，使得本系統具有更高的彈性、效能、容錯能力與擴充性，來提供更具效能之網際網路系統線上服務。
關鍵字: 慢性腎臟病;人工智慧;模糊專家系統;風險評估;雲端運算服務;
The Study of Cloud Service Information System - A Case Study of Online Detection and Risk Assessment for Chronic Kidney Disease
Abstract: Purpose - By the annual mortality statistics announced by the Department of Health Welfare in year 105, nephritis, nephrotic syndrome and nephrosis are ranked among the top ten leading causes of civilian death. Again, according to the 2017 Annual Data Report of United States Renal Data System, it indicated that the mortality rate in end stage renal disease (ESRD) in Taiwan was ranked in the first place in the world. In general, it is difficult to be detected because the symptoms of chronic kidney disease (CKD) are not very apparent. Once ESRD is detected, it may require to rely on the dialysis to maintain alive. Design/methodology/approach - Firstly, through the literature review and interview from experts, the methods of detection and risk factors of chronic kidney disease are determined. Secondly, leverage the factors identified, two artificial intelligent techniques including the back-propagation network (BPN) and support vector machines (SVM) are built and employed to verify the accuracy of these factors of affecting CKD. Lastly, Glomerular Filtration Rate (GFR) is employed to detect whether a system user has CKD followed by the fuzzy expert system is employed to assess the deterioration of probable risk. Findings - Through the aid of this service system, the publics may take advantage of this system over the Internet to conduct an online self-assessment for the risk of suffering CKD. In the meanwhile, the system is deployed on the Google cloud platform to provide a higher performance system online service. Research limitations/implications - Because the volume of patient cases employed in this research are not large enough, the reliability of research outcome may be questioned by the specialty physicians in medical practice even the predication model is thoroughly built and tested. If the source of patient cases can be obtained and extracted from National Health Insurance Research Database (NHIRD), the reliability of outcome would be enhanced to be trusted. Practical implications - The verification and validation of the this research regardless the methods used, model built, and outcomes generated should be furtherly done in order to put the result in practice in medicine domain. In additions, the powerful computing and dynamical adjustment of cloud infrastructure needs also to be cared to make the service system gaining higher flexibility, better performance, greater fault-tolerance and scalability to improve the performance of system online service over the Internet. Originality/value - The early diagnosis and treatment of CKD is extremely important. The study of this paper builds a cloud information service system based on the techniques of cloud computing, fuzzy expert system, and system development tool, jFuzzyLogic The participation of the kidney physician in this research provides the valuable and expert knowledge to build the original and unique predicting model for this research. If the research outcome and methodologies can be furtherly verified and validated, the value of academics and practices of this research can be also expected.
Keywords: chronic kidney disease;artificial intelligence;fuzzy expert system;risk assessment;cloud computing service;
瀏覽次數: 17798 下載次數: 183
引用 導入Endnote