臺灣博碩士論文加值系統

近年來因心血管疾病而死亡的人數越來越多，是一個不能輕忽的疾病。心血管疾病在發病前，沒有特別明顯的症狀，常常被忽略。臨床上，醫生可以透過運動心電圖來檢查病人是否有心血管疾病。運動心電圖是讓病人於跑步機上行走，記錄其運動狀態的心電圖及血壓和心跳數。文獻上，醫生利用運動心電圖診斷心血管疾病的準確率大約落在70~75%。在本論文中，研究使用各種機器學習分類演算法進行運動心電圖數據的分析。實驗數據是100名懷疑和已知冠狀動脈疾病的患者數據，這些患者經過醫生診斷後接受運動心電圖檢查。本研究使用五種機器學習分類算法:最近鄰居法、決策樹、隨機森林、支持向量機和極限梯度提升。實驗結果，極限梯度提升分類器在所有分類模型具有最高的性能表現，可以達到84%的正確率。結果顯明，機器學習方法可以用來輔助醫生提高心血管疾病診斷的準確率。

In recent years, the number of people dying from cardiovascular diseases is increasing, and it is a disease that cannot be ignored. Cardiovascular disease has no particularly obvious symptoms before onset and is often overlooked. Clinically, doctors can check whether a patient has cardiovascular disease through a sports ECG. The exercise ECG is an electrocardiogram and blood pressure and heart rate that allow the patient to walk on the treadmill and record their movement status. In the literature, the accuracy of doctors using cardiovascular electrocardiography to diagnose cardiovascular disease falls to about 70-75%. In this paper, we study the use of various machine learning classification algorithms for the analysis of exercise ECG data. The experimental data is data from 100 patients with suspected and known coronary artery disease who underwent exercise electrocardiography after a doctor's diagnosis. We use five machine learning classification algorithms: K-Nearest Neighbor (KNN), Decision Tree (DT), Random Forest(RF), Support Vector Machine(SVM) and eXtreme Gradient Boosting (XGBboost). As a result of the experiment, the XGBboost classifier has the highest performance in all classification models and can achieve an accuracy rate of 84%. The results show that machine learning methods can be used to assist doctors in improving the accuracy of cardiovascular disease diagnosis.

摘要 I
ABSTRACT II
目錄 III
圖目錄 IV
表目錄 I
第一章緒論 1
1.1研究背景與動機 1
1.2研究目的 1
1.3論文架構 2
第二章文獻探討 3
2.1心血管疾病 3
2.2運動心電圖 3
2.3機器學習 5
第三章實驗方法與結果 8
3.1實驗流程 8
3.2運動心電圖資料 8
3.3資料預處理 11
3.4特徵選擇 13
3.5訓練模型 13
3.6參數最佳化 20
3.7驗證模型 27
3.8實驗結果 28
第四章結論與未來展望 32
4.1結論 32
4.2未來展望 32
參考文獻 34
圖1:論文架構 2
圖2:運動心電圖檢查儀器 4
圖3:模型建構架構 6
圖4:實驗流程圖 8
圖5:運動心電圖檢查數據 9
圖6:運動心電波型圖 10
圖7:最近鄰居法 14
圖8:決策樹 16
圖9:隨機森林 17
圖10:支持向量機 18
圖11:參數最佳化流程圖 21
圖12:最近鄰居參數最佳化結果 23
圖13:隨機森林參數最佳化結果 23
圖14:決策樹參數最佳化結果 24
圖15:支持向量機參數最佳化結果 24
圖16:極限梯度提升參數最佳化結果 27
圖17:特徵評分 30
圖18:各模型結果比較 31
表格1:運動心電圖基本資訊 10
表格2:特徵分級範圍表 12
表格3:各分類模型使用之參數 22
表格4:使用不同K值的最近鄰居模型 28
表格5:使用最大決策樹個數k及樹深度k的隨機森林模型 28
表格6:使用不同核心及懲罰係數的支持向量機 29
表格7:未參數最佳化的模型比較 30
表格8:參數最佳化後的各模型比較 31

1.Al'Aref, S.J., K. Anchouche, G. Singh, P.J. Slomka, K.K. Kolli, A. Kumar, M. Pandey, G. Maliakal, A.R. van Rosendael, A.N. Beecy, D.S. Berman, J. Leipsic, K. Nieman, D. Andreini, G. Pontone, U.J. Schoepf, L.J. Shaw, H.J. Chang, J. Narula, J.J. Bax, Y. Guan, and J.K. Min, Clinical applications of machine learning in cardiovascular disease and its relevance to cardiac imaging. Eur Heart J, 2019. 40(24): p. 1975-1986.
2.黃瑞仁, 冠狀動脈心臟病的診斷與治療. 警政署日新雜誌, vol, 2004(3): p. 208-215.
3.Mendis, S., P. Puska, B. Norrving, and W.H. Organization, Global atlas on cardiovascular disease prevention and control. 2011: Geneva: World Health Organization.
4.Fares, A., Winter cardiovascular diseases phenomenon. North American journal of medical sciences, 2013. 5(4): p. 266-279.
5.Vilcant, V. and R. Zeltser, Treadmill Stress Testing, in StatPearls. 2019: Treasure Island (FL).
6.Fihn, S.D., J.C. Blankenship, K.P. Alexander, J.A. Bittl, J.G. Byrne, B.J. Fletcher, G.C. Fonarow, R.A. Lange, G.N. Levine, T.M. Maddox, S.S. Naidu, E.M. Ohman, and P.K. Smith, 2014 ACC/AHA/AATS/PCNA/SCAI/STS focused update of the guideline for the diagnosis and management of patients with stable ischemic heart disease: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines, and the American Association for Thoracic Surgery, Preventive Cardiovascular Nurses Association, Society for Cardiovascular Angiography and Interventions, and Society of Thoracic Surgeons. J Am Coll Cardiol, 2014. 64(18): p. 1929-49.
7.Turing, A.M., Computing machinery and intelligence, in Parsing the Turing Test. 2009, Springer. p. 23-65.
8.Ramalingam, V.V., A. Dandapath, and M. Karthik Raja, Heart disease prediction using machine learning techniques: A survey. Vol. 7. 2018. 684.
9.陳志華, 楊子緯, 張訓楨, and 賴永崧, 特徵分析和機器學習方法應用於肝臟疾病檢測. 福祉科技與服務管理學刊, 2016. 4(3): p. 417-429.
10.Swaminathan, S., K. Qirko, T. Smith, E. Corcoran, N.G. Wysham, G. Bazaz, G. Kappel, and A.N. Gerber, A machine learning approach to triaging patients with chronic obstructive pulmonary disease. PloS one, 2017. 12(11): p. e0188532-e0188532.
11.Ali, E.E.E. and W.Z. Feng, Breast Cancer Classification using Support Vector Machine and Neural Network. International Journal of Science and Research, 2016. 5(3): p. 1-6.
12.Okfalisa, I. Gazalba, Mustakim, and N.G.I. Reza. Comparative analysis of k-nearest neighbor and modified k-nearest neighbor algorithm for data classification. in 2017 2nd International conferences on Information Technology, Information Systems and Electrical Engineering (ICITISEE). 2017.
13.Gavankar, S.S. and S.D. Sawarkar. Eager decision tree. in 2017 2nd International Conference for Convergence in Technology (I2CT). 2017.
14.Sakr, S., R. Elshawi, A.M. Ahmed, W.T. Qureshi, C.A. Brawner, S.J. Keteyian, M.J. Blaha, and M.H. Al-Mallah, Comparison of machine learning techniques to predict all-cause mortality using fitness data: the Henry ford exercIse testing (FIT) project. BMC Med Inform Decis Mak, 2017. 17(1): p. 174.
15.An, G., K. Omodaka, S. Tsuda, Y. Shiga, N. Takada, T. Kikawa, T. Nakazawa, H. Yokota, and M. Akiba, Comparison of Machine-Learning Classification Models for Glaucoma Management. J Healthc Eng, 2018. 2018: p. 6874765.
16.Chen, T. and C. Guestrin, XGBoost. 2016: p. 785-794.
17.Zaman, M.u., N. Fatima, U. Zaman, and D.J. Baloch, High negative predictive value of workload ≥7 METS on exercise testing in patients with normal gated myocardial perfusion imaging: Was imaging really required? Iranian Journal of Nuclear Medicine, 2014. 22(2): p. 70-76.
18.Sharif, S. and S. Alway, The diagnostic value of exercise stress testing for cardiovascular disease is more than just st segment changes: a review. J. Integr. Cardiol, 2016. 2(4): p. 41-55.
19.Yi-Min, H. and D. Shu-Xin. Weighted support vector machine for classification with uneven training class sizes. in 2005 International Conference on Machine Learning and Cybernetics. 2005.
20.Li, J., C. Zhang, and Z. Li. Battlefield Target Identification Based on Improved Grid-Search SVM Classifier. in 2009 International Conference on Computational Intelligence and Software Engineering. 2009.
21.Gao, X., S. Fan, X. Li, Z. Guo, H. Zhang, Y. Peng, and X. Diao. An improved XGBoost based on weighted column subsampling for object classification. in 2017 4th International Conference on Systems and Informatics (ICSAI). 2017.
22.Sanders, S. and C. Giraud-Carrier, Informing the Use of Hyperparameter Optimization Through Metalearning. 2017: p. 1051-1056.
23.Duan, G. and X. Ma. A Coupon Usage Prediction Algorithm Based On XGBoost. in 2018 14th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-FSKD). 2018.
24.Elaidi, H., Y. Elhaddar, Z. Benabbou, and H. Abbar. An idea of a clustering algorithm using support vector machines based on binary decision tree. in 2018 International Conference on Intelligent Systems and Computer Vision (ISCV). 2018.
25.Pławiak, P., Novel genetic ensembles of classifiers applied to myocardium dysfunction recognition based on ECG signals. Swarm and Evolutionary Computation, 2018. 39: p. 192-208.
26.Zhang, X., Y. Yang, and Z. Zhou. A novel credit scoring model based on optimized random forest. in 2018 IEEE 8th Annual Computing and Communication Workshop and Conference (CCWC). 2018.
27.Shi, H., H. Wang, Y. Huang, L. Zhao, C. Qin, and C. Liu, A hierarchical method based on weighted extreme gradient boosting in ECG heartbeat classification. Comput Methods Programs Biomed, 2019. 171: p. 1-10.
28.Bergstra, J. and Y. Bengio, Random search for hyper-parameter optimization. J. Mach. Learn. Res., 2012. 13: p. 281-305.
29.Yadav, S. and S. Shukla. Analysis of k-Fold Cross-Validation over Hold-Out Validation on Colossal Datasets for Quality Classification. in 2016 IEEE 6th International Conference on Advanced Computing (IACC). 2016.