台灣的末期腎臟疾病與血液透析盛行率是世界第一，因腎炎、腎病症候群及腎
病變死亡在台灣十大死亡原因中排名第九，而急性腎損傷(Acute Kidney Injury, AKI)
是臨床上常見且複雜的症狀，急性腎損傷是健康的腎臟突然受到某種因素的傷害，
而使腎臟的排除毒素的功能降低，無法正常的運作。造成急性腎損傷的因素有慢性
腎臟疾病、藥物間的交互作用或是服用腎毒性藥物所導致，例如非類固醇消炎藥
(NSAIDs)會抑制前列腺素使腎絲球血流量降低，進而使腎絲球過濾率下降，而導致
腎臟損傷。因此，若能事先預防或是早期發現早期治療，避免產生不必要的併發症，
可提高腎臟恢復的機會，提高病患的存活率以及降低醫療成本。而藥物間交互作用
也是導致急性腎損傷的因素之一，藥物間交互作用是指當一種藥物與另一種藥物同
時服用時，藥物對身體的影響發生變化，可能產生延遲，減少或增強任一種藥物的
吸收，而引起不良反應，例如被稱之為腎臟的”三重打擊”的非類固醇消炎藥、利尿
劑(Diuretics)和血管收縮素轉換酶抑製劑(ACE)/血管張力素受體阻滯劑(ARA)三種
藥物同時服用對腎臟的傷害。我們透過台灣健保資料庫中，曾經患有急性腎損傷的
病患的用藥紀錄，去發掘更多的藥物與疾病的組合，以及導致 AKI 的序列發現，我
們使用我們開發的機器學習的演算法，深度規則森林（Deep Rule Forests, DRF），
幫助我們從樹的模型中發現和提取規則，我們發現深度規則森林所找出的藥物與疾
病的組合，與文獻中相符合，也找出文獻中尚未發掘的組合，準確度方面也較傳統
單一樹模型好。而交互作用序列偵測我們使用了 DRF 結合隱半馬可夫模型(Hidden
Semi-Markov Model, HSMM)，它幫助我們找出藥物與疾病導致急性腎損傷的序列組
合，其所找出的規則與文獻中也相符合，也找出了文獻中尚未發現的序列組合，證
明了 DRF 與 HSMM 能幫助我們得到更精準的規則與序列。

The prevalence rate of end stage renal disease and hemodialysis in Taiwan is the highest in the world. Death due to nephritis and kidney disease ranks the 9th among the top ten leading causes of death in Taiwan. Acute Kidney Injury (AKI) is a common and complex symptom in clinical practice. AKI means a sudden damage of a healthy kidney caused by certain factors, making the detoxification of the kidney decline and not work properly. The factors that cause AKI include chronic kidney disease, drug-drug interactions and nephrotoxic drugs. For example, Non-Steroidal Anti-Inflammatory Drugs (NSAIDs) inhibit prostaglandins, which causes glomerular blood flow to decrease; afterwards, the glomerular filtration rate has declined, leading to the damage of the kidney.
Therefore, unnecessary complications can be avoided if any symptoms are diagnosed and treated at an early stage. It will increase chances of kidney recovery, improve survival rates and lower the medical costs. In addition, drug-drug interactions are also one of the factors leading to AKI. Drug-drug interactions means two or more drugs are taken at the same time and one of the drugs may cause the absorption of another drug to be delayed, reduced or increased. For example, NSAIDs, Diuretics and Angiotensin converting enzyme (ACE)/ Angiotensin II Receptor Blockers (ARA) are called "triple whammy" for the kidney. If they are taken at the same time, there will be damage to the kidney.
By using National Health Insurance (NHI) database in Taiwan, we discovered more combinations of drugs and diseases through the medication record of patients with AKI and found out the order of taking medicine that resulted in AKI. We used Deep Rule Forests (DRF), the machine learning algorithm, which we developed, to help us discover and get rules from the tree model. We found out the combinations of drugs and diseases by DRF are consistent with the literature. It also found out the combinations that have not been explored in the literature. The accuracy is higher than the traditional single tree model. As for the detection of Drug-Disease Interaction, we used DRF combined with Hidden SemiMarkov Model (HSMM). It can help us to discover the combinations of drugs and diseaseinduced AKI, and the rules are consistent with the literature. Likewise, it also found out the combinations that have not been mentioned in the literature. Therefore, DRF and HSMM can help us obtain more precise rules and sequences.

致謝 ii
摘要 iii
Abstract iv
目錄 vi
第 一 章 研究背景、動機、目的 1
1.1 研究背景 1
1.2 研究動機 2
1.3 研究目的 3
第 二 章 文獻探討 3
2.1 急性腎損傷(Acute Kidney Injury, AKI) 3
2.2 慢性腎臟疾病(Chronic Kidney Disease, CKD) 5
2.3 藥物引發腎毒性 7
2.4 藥物與疾病相互作用(Drug-Disease Interaction) 10
2.5 藥物相互作用(Drug-Drug Interaction, DDI) 10
第 三 章 研究方法 12
3.1 Tree-based method 12
3.2 深度規則森林 (Deep Rule Forests, DRF) 13
3.3 Elastic net 14
3.4 Hidden Semi-Markov Models 15
3.5 Classification Tree Hidden Semi-Markov Model (CTHSMM) 17
第 四 章 研究結果 20
4.1 資料來源 20
4.2 研究結果 23
4.2.1 模型比較 24
4.2.2 規則比較 27
4.2.3 序列發現 30
第 五 章 結論 38
第 六 章 參考文獻 39

Ali, T., Khan, I., Simpson, W., Prescott, G., Townend, J., Smith, W., & MacLeod, A. (2007). Incidence and Outcomes in Acute Kidney Injury: A Comprehensive Population-Based Study. Journal of the American Society of Nephrology, 18(4), 1292–1298.
Bagshaw, S. M., George, C., & Bellomo, R. (2008). Early acute kidney injury and sepsis: a multicentre evaluation. Critical Care, 12(2), R47.
Basile, D. P., Anderson, M. D., & Sutton, T. A. (2012). Pathophysiology of Acute Kidney Injury. In R. Terjung (Ed.), Comprehensive Physiology. Hoboken, NJ, USA: John Wiley & Sons, Inc.
Belayev, L. Y., & Palevsky, P. M. (2014). The link between acute kidney injury and chronic kidney disease: Current Opinion in Nephrology and Hypertension, 23(2), 149–154.
Bellomo, R., Ronco, C., Kellum, J. A., Mehta, R. L., & Palevsky, P. (2004). Acute renal failure – definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group, 8(4), 9.
Blantz, R. C. (1998). Pathophysiology of pre-renal azotemia. Kidney International, 53(2), 512–523.
Common Drug-Disease Interactions. (2003). Retrieved December 8, 2018, from http://study.com/academy/lesson/common-drug-disease-interactions.html
Drug-Drug Interaction Definition. (2018). Retrieved October 17, 2018, from https://aidsinfo.nih.gov/understanding-hiv-aids/glossary/213/drug-druginteraction
Girardeau, Y., Trivin, C., Durieux, P., Le Beller, C., Louet Agnes, L.-L., Neuraz, A., … Avillach, P. (2015). Detection of Drug–Drug Interactions Inducing Acute Kidney Injury by Electronic Health Records Mining. Drug Safety, 38(9), 799–809.
Hassan, Y., Al-Ramahi, R., Aziz, N. A., & Ghazali, R. (2009). Drug Use and Dosing in Chronic Kidney Disease, 38(12), 10.
Hegarty, N. J., Young, L. S., Kirwan, C. N., O’Neill, A. J., Bouchier-Hayes, D. M., Sweeney, P., … Fitzpatrick, J. M. (2001). Nitric oxide in unilateral ureteral obstruction: Effect on regional renal blood flow. Kidney International, 59(3),1059–1065.
Hsu, C. -y., McCulloch, C. E., Fan, D., Ordoñez, J. D., Chertow, G. M., & Go, A. S. (2007). Community-based incidence of acute renal failure. Kidney International, 72(2), 208–212.
Inker, L. A., Levey, A. S., Pandya, K., Stoycheff, N., Okparavero, A., & Greene, T. (2014). Early Change in Proteinuria as a Surrogate End Point for Kidney Disease Progression: An Individual Patient Meta-analysis. American Journal of Kidney Diseases, 64(1), 74–85.
James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An Introduction to Statistical Learning (Vol. 103). New York, NY: Springer New York.
Kang, Y., & Zadorozhny, V. (2016). Process Discovery Using Classification Tree Hidden Semi-Markov Model. In 2016 IEEE 17th International Conference on Information Reuse and Integration (IRI) (pp. 361–368). Pittsburgh, PA, USA:IEEE.
Kaufman, J., Dhakal, M., Patel, B., & Hamburger, R. (1991). Community-Acquired Acute Renal Failure. American Journal of Kidney Diseases, 17(2), 191–198.
Kellum, J. A. (2008). Defining and classifying AKI: one set of criteria. Nephrology Dialysis Transplantation, 23(5), 1471–1472.
Lameire, N., Biesen, W. V., & Vanholder, R. (2008). Acute kidney injury, 372, 3. Landry, M., & Bartz, A. (2018). Machine Learning with R and H2O, 46.
Lapi, F., Azoulay, L., Yin, H., Nessim, S. J., & Suissa, S. (2013). Concurrent use of diuretics, angiotensin converting enzyme inhibitors, and angiotensin receptor blockers with non-steroidal anti-inflammatory drugs and risk of acute kidney injury: nested case-control study. BMJ, 346(jan08 12), e8525–e8525.
Levey, A. S., & Coresh, J. (2012). Chronic kidney disease. The Lancet, 379(9811), 165–180.
Makris, K., & Spanou, L. (2016). Acute Kidney Injury: Definition, Pathophysiology and Clinical Phenotypes. Acute Kidney Injury, 14.
Mehta, R. L., Kellum, J. A., Shah, S. V., Molitoris, B. A., Ronco, C., Warnock, D. G., & Levin, A. (2007). Acute Kidney Injury Network: report of an initiative to improve outcomes in acute kidney injury. Critical Care, 11(2), R31.
Naughton, C. A. (2008). Drug-Induced Nephrotoxicity, 78(6), 8.
Oates, J. A. (1969). Drug-drug interaction. Interference with the delivery of drugs to their sites of action. JAMA: The Journal of the American Medical Association,208(10), 1898b – 1898.
O’Connell, J., & Højsgaard, S. (2011). Hidden Semi Markov Models for Multiple Observation Sequences: The mhsmm Package for R. Journal of Statistical Software, 39(4).
Prieto-García, L., Pericacho, M., Sancho-Martínez, S. M., Sánchez, Á ., MartínezSalgado, C., López-Novoa, J. M., & López-Hernández, F. J. (2016). Mechanisms of triple whammy acute kidney injury. Pharmacology & Therapeutics, 167, 132–145.
Qato, D. M., Wilder, J., Schumm, L. P., Gillet, V., & Alexander, G. C. (2016). Changes in Prescription and Over-the-Counter Medication and Dietary Supplement Use Among Older Adults in the United States, 2005 vs 2011. JAMA Internal Medicine, 176(4), 473.
Ricci, Z., Cruz, D. N., & Ronco, C. (2011). Classification and staging of acute kidney injury: beyond the RIFLE and AKIN criteria. Nature Reviews Nephrology, 7(4), 201–208.
Rosas-Carrasco, Ó ., García-Peña, C., Sánchez-García, S., Vargas-Alarcón, G., Gutiérrez-Robledo, L. M., & Juárez-Cedillo, T. (2011). The relationship between potential drug-drug interactions and mortality rate of elderly hospitalized patients, 10.
Schetz, M., Dasta, J., Goldstein, S., & Golper, T. (2005). Drug-induced acute kidney injury, 11.
Stevens, L. A., Coresh, J., Greene, T., & Levey, A. S. (2006). Assessing Kidney Function — Measured and Estimated Glomerular Filtration Rate. New England Journal of Medicine, 354(23), 2473–2483.
Therneau, T. M., Atkinson, E. J., & Foundation, M. (2018). An Introduction to Recursive Partitioning Using the RPART Routines, 62. USRDS. (2016). Acute Kidney Injury. ACUTE KIDNEY INJURY, 1, 26.
Webster, A. C., Nagler, E. V., Morton, R. L., & Masson, P. (2017). Chronic Kidney Disease. The Lancet, 389(10075), 1238–1252.
Wright, M. N., & Ziegler, A. (2017). ranger: A Fast Implementation of Random Forests for High Dimensional Data in C++ and R. Journal of Statistical Software, 77(1).
Yu, S.-Z. (2010). Hidden semi-Markov models. Artificial Intelligence, 174(2), 215–243.
Zhang, S., Lu, Y., Zhang, L., Cai, B., & Qiu, K. (2015). Variable Selection in Logistic Regression Model. Chinese Journal of Electronics, 24(4), 813–817.
衛生福利部中央健康保險署. (2016, October 1). 衛生福利部中央健康保險署 [文字]. Retrieved January 16, 2019, from https://www.nhi.gov.tw/Content_List.aspx?n=B5612D76EB95D83D&topn=CA428784F9ED78C9&Create=1
衛生福利部國民健康署. (2016, December 31). 衛生福利部國民健康署 [text/html].Retrieved November 17, 2018, from https://www.hpa.gov.tw/Pages/Detail.aspx?nodeid=635&pid=1200