論文使用權限 Thesis access permission:自定論文開放時間 user define
開放時間 Available:
校內 Campus: 已公開 available
校外 Off-campus: 已公開 available
論文名稱 Title |
基於深度多工學習演算法的共病症特徵學習-以腎臟疾病為例 Renal Comorbidity Discovery based on Deep Multitasking Learning |
||
系所名稱 Department |
|||
畢業學年期 Year, semester |
語文別 Language |
||
學位類別 Degree |
頁數 Number of pages |
60 |
|
研究生 Author |
|||
指導教授 Advisor |
|||
召集委員 Convenor |
|||
口試委員 Advisory Committee |
|||
口試日期 Date of Exam |
2019-01-25 |
繳交日期 Date of Submission |
2019-01-30 |
關鍵字 Keywords |
機器學習、急性腎損傷、慢性腎臟病、深度多工學習、深度規則森林 Deep Multitasking Learning, DRF, AKI, CKD, Machine Learning |
||
統計 Statistics |
本論文已被瀏覽 6186 次,被下載 2 次 The thesis/dissertation has been browsed 6186 times, has been downloaded 2 times. |
中文摘要 |
腎臟病在台灣一直是名列十大的死因之一,期造成原因為老年人口的增加,肥胖、慢性疾病、心血管疾病病人也逐漸提升,另外,根據健保局資料統計,國人濫用藥物習慣也逐漸偏高,若沒有經過醫師指示,藥物本身存在的腎毒性也是造成腎臟疾病的主要原因,已有專家證實含有腎毒性的藥物的西藥約有150種以上,中藥約有50種以上,另外,共病特徵影響也是造成死亡率的關鍵因素,在健保資料統計中65歲以上老人自述經醫生診斷後確定罹患慢性疾病數的百分比,一項以上為88.71%,兩項以上為71.67%,三項以上為51.25%,此趨勢逐漸證明當多種疾病存在於人體中的時候,疾病之間的關聯也是我們需注意的方向之一,例如目前已證實的糖尿病、癌症、心臟手術、人類免疫缺陷病毒(HIV)獲得性免疫缺陷綜合症(艾滋病)和癌症都是急性腎損傷的共病症關鍵因子,因此本研究的目的即是找出當存在急性腎損傷及慢性腎衰竭時的共病特徵有哪些,而在研究中我們提出了一種學習方法,即深度規則森林(Deep Rule Forest, DRF),此方法的方式透過逐層不斷的重複訓練,再從訓練中去萃取其最佳規則,這當中本研究也嘗試從單一疾病的觀點即合併疾病的觀點來分析共病症,在分析這些規則中的疾病也符合了文獻及專家的研究和佐證,「DRF」的方法確實也幫助本研究在模型觀察指標上的提升,我們也因此證明層數越深的訓練,最後的結果也會比準確。 |
Abstract |
Kidney disease has been one of the top ten causes of death in Taiwan. The increasing population of the elderly, and patients with obesity, chronic diseases and cardiovascular diseases are the major causes of the diseases. According to National Health Insurance Administration Ministry of Health and Welfare drug abuse in Taiwan has been rising. Without doctor’s instructions and prescription, Nephrotoxicity of the drug itself can be a major cause of kidney disease. Furthermore, experts have confirmed approximately 150 kinds of western medicines and 50 kinds of Chinese medicines contain nephrotoxic drugs. In addition, Comorbidity is also a critical factor in mortality. The percentage of people over the age of 65 who have been diagnosed with one chronic disease is 88.71%, with two and three or more are 71.67% and 51.25% respectively. The statistics prove that when multiple diseases exist in the body, the correlations among diseases are also one of the directions we need to research. For example, diabetes, cancer, heart disease, human immunodeficiency virus (HIV) acquired immunodeficiency syndrome (AIDS) and cancer have all been proved to be key factors in comorbidities in acute kidney injury. Therefore, the purpose of this research is to identify comorbidities in acute kidney injury and chronic renal failure. We proposed a learning algorithm called Deep Rule Forest (DRF), which can find the best rule by layer-by-layer repeated training. In this research, we also analyzed the comorbidities in the view of one single disease and combined diseases which confirmed the literatures and the research of experts. The DRF method does help the research to improve the model observation indicators. As the result, we prove that if the layers of DRF model are deeper, the extracted rules are more precise. |
目次 Table of Contents |
論文審訂書 i Abstract ii 摘要 iii 致謝 iv 圖目錄 vi 表目錄 vii 第一章 緒論 1 1.1 研究背景 1 1.2 研究動機 3 1.3 研究目的 5 第二章 文獻探討 6 2.1 共病症或合併症 6 2.2 腎臟病學(Nephrology) 9 2.3急性腎損傷Acute Kidney Injury (AKI) 10 2.4 慢性腎衰竭Chronic Kidney Disease (CKD) 13 2.5 AKI發生率增加的相關主要共病疾病 16 第三章 研究方法與步驟 18 3.1 研究方法 18 3.1.1 LASSO 20 3.1.2 ELASTIC NET 22 3.1.3 決策樹 24 3.1.4 深度規則森林(Deep Rule Forest, DRF) 27 3.2 研究架構 30 3.3 評估標準 32 第四章 研究結果與分析 33 4.1資料整理 33 4.2 研究流程 35 4.3 研究過程 36 4.3.1 Raw data—Predict AKI 37 4.3.2 Raw data—Predict CKD 39 4.3.3 DRF (Deep Rule Forest)—Predict AKI 41 4.3.4 DRF—Merge AKI & CKD Predict AKI 43 4.4 研究分析 46 第五章 研究結論與建議 49 5.1 研究結論 49 第六章 文獻參考 51 |
參考文獻 References |
Bennet, S. J., Berry, O. M., Goddard, J., & Keating, J. F. (2010). Acute renal dysfunction following hip fracture. Injury, 41(4), 335-338. doi:10.1016/j.injury.2009.07.009 Brown, J. R., MacKenzie, T. A., Maddox, T. M., Fly, J., Tsai, T. T., Plomondon, M. E., . . . Matheny, M. E. (2015). Acute Kidney Injury Risk Prediction in Patients Undergoing Coronary Angiography in a National Veterans Health Administration Cohort With External Validation. J Am Heart Assoc, 4(12). doi:10.1161/JAHA.115.002136 Chao, C. T., Wang, J., Wu, H. Y., Huang, J. W., & Chien, K. L. (2018). Age modifies the risk factor profiles for acute kidney injury among recently diagnosed type 2 diabetic patients: a population-based study. Geroscience, 40(2), 201-217. doi:10.1007/s11357-018-0013-3 Charlson, M. E., Pompei, P., Ales, K. L., & MacKenzie, C. R. (1987). A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis, 40(5), 373-383. Chawla, L. S., Eggers, P. W., Star, R. A., & Kimmel, P. L. (2014). Acute kidney injury and chronic kidney disease as interconnected syndromes. N Engl J Med, 371(1), 58-66. doi:10.1056/NEJMra1214243 Christiansen, C. F., Johansen, M. B., Langeberg, W. J., Fryzek, J. P., & Sorensen, H. T. (2011). Incidence of acute kidney injury in cancer patients: a Danish population-based cohort study. Eur J Intern Med, 22(4), 399-406. doi:10.1016/j.ejim.2011.05.005 Efron, e. a. (2004). LEAST ANGLE REGRESSION. Farooqi, S., & Dickhout, J. G. (2016). Major comorbid disease processes associated with increased incidence of acute kidney injury. World J Nephrol, 5(2), 139-146. doi:10.5527/wjn.v5.i2.139 Fonti, V. (2017). Feature selection using LASSO. Vrije Universiteit Amsterdam. Golub, T. R., Slonim, D. K., Tamayo, P., Huard, C., Gaasenbeek, M., Mesirov, J. P., . . . Lander, E. S. (1999). Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science, 286(5439), 531-537. Hobson, C., Ozrazgat-Baslanti, T., Kuxhausen, A., Thottakkara, P., Efron, P. A., Moore, F. A., . . . Bihorac, A. (2015). Cost and Mortality Associated With Postoperative Acute Kidney Injury. Ann Surg, 261(6), 1207-1214. doi:10.1097/sla.0000000000000732 Hui Zou, T. H. (2003). Regression Shrinkage and Selection via the Elastic Net, with Applications to Microarrays. Inker, L. A., Astor, B. C., Fox, C. H., Isakova, T., Lash, J. P., Peralta, C. A., . . . Feldman, H. I. (2014). KDOQI US commentary on the 2012 KDIGO clinical practice guideline for the evaluation and management of CKD. Am J Kidney Dis, 63(5), 713-735. doi:10.1053/j.ajkd.2014.01.416 Kim, S. M., Kim, Y., Jeong, K., Jeong, H., & Kim, J. (2018). Logistic LASSO regression for the diagnosis of breast cancer using clinical demographic data and the BI-RADS lexicon for ultrasonography. Ultrasonography, 37(1), 36-42. doi:10.14366/usg.16045 Lameire, N. (2014). Nephrotoxicity of recent anti-cancer agents. Clin Kidney J, 7(1), 11-22. doi:10.1093/ckj/sft135 Lopes, J. A., & Jorge, S. (2013). The RIFLE and AKIN classifications for acute kidney injury: a critical and comprehensive review. Clinical Kidney Journal, 6(1), 8-14. Stamey, T. A., Kabalin, J. N., McNeal, J. E., Johnstone, I. M., Freiha, F., Redwine, E. A., & Yang, N. (1989). Prostate specific antigen in the diagnosis and treatment of adenocarcinoma of the prostate. II. Radical prostatectomy treated patients. J Urol, 141(5), 1076-1083. Szczech, L. A., Granger, C. B., Dasta, J. F., Amin, A., Peacock, W. F., McCullough, P. A., . . . Studying the Treatment of Acute Hypertension, I. (2010). Acute kidney injury and cardiovascular outcomes in acute severe hypertension. Circulation, 121(20), 2183-2191. doi:10.1161/CIRCULATIONAHA.109.896597 USRDS. (2018). 2017 USRDS Annual Data Report: Executive Summary. American Journal of Kidney Diseases, 71(3), S1-S8. doi:10.1053/j.ajkd.2018.01.003 Xianzhen Zhang, H. W. (2017). A study on the risk factors for acute kidney injury in acute gouty arthritis. Yoshua Bengio, O. D., Clarence Simard. (2007). Decision Trees do not Generalize to New. Zhi-Hua Zhou, J. F. Deep Forest. 內政部統計處. (2017). 我國老年人口數首次超過幼年人口數. 台灣腎臟學會. <慢性腎臟病流行病學>. 衛生福利部. (2017). 105年國民醫療保健支出. 衛生福利部國民健康署. (2017). 2017台灣腎病年報. |
電子全文 Fulltext |
本電子全文僅授權使用者為學術研究之目的,進行個人非營利性質之檢索、閱讀、列印。請遵守中華民國著作權法之相關規定,切勿任意重製、散佈、改作、轉貼、播送,以免觸法。 論文使用權限 Thesis access permission:自定論文開放時間 user define 開放時間 Available: 校內 Campus: 已公開 available 校外 Off-campus: 已公開 available |
紙本論文 Printed copies |
紙本論文的公開資訊在102學年度以後相對較為完整。如果需要查詢101學年度以前的紙本論文公開資訊,請聯繫圖資處紙本論文服務櫃台。如有不便之處敬請見諒。 開放時間 available 已公開 available |
QR Code |