高血壓是心血管疾病的主要危險因素，左心室肥大是高血壓常見的併發症之一，也是心血管事件發生的一項獨立危險因子。傳統臨床診斷左心室肥大的方式有心電圖和超音波，心電圖的診斷標準有許多種，臨床醫師一般會再經由心臟超音波進一步確認。本論文採用機器學習提高預測效果，九個生理指數為機器學習模型之輸入，機器學習的四種模型採用邏輯回歸、決策樹、梯度提升決策樹和貝葉斯分類器，評估標準有準確度、敏感度、特異度、查準率、F_1分數、ROC曲線之AUC以及PR曲線之AUC。

Hypertension is one of the main risk factors of cardiovascular disease (CVD). Left ventricular hypertrophy (LVH) is a common complication of hypertension and also an independent risk factor of CVD events. Clinically, the traditional diagnosis of LVH depends on electrocardiogram (ECG) and echocardiography. Many ECG criteria for diagnosing LVH have been established. The clinicians may use echocardiography for a further confirmation of ECG-based LVH. This thesis uses machine learning to improve the prediction performance. Nine physiological indices are the inputs to machine learning models. The four models of machine learning include logistic regression, decision tree, gradient boost decision tree and Bayes' classifier. The evaluation criteria are accuracy, sensitivity, specificity, precision, F_1 score, the area under the curve of receiver operating characteristic (AUC ROC), and the area under the curve of precision recall (AUC PR).

摘要......................1
Abstract..................3
目錄......................5
表目錄....................8
圖目錄....................9
第一章 緒論...............11
1.1左心室肥大介紹.......11
1.2人工智慧之概述.......13
1.3研究動機............17
1.4論文架構............18
第二章 相關文獻回顧.......19
2.1傳統左心室肥大檢測標準與文獻回顧........24
2.2機器學習方法探討......................27
2.2.1泛化能力(Generalization Ability)....27
2.2.2邏輯回歸(Logistic Regression).......29
2.2.3決策樹(Decision Tree)...............32
2.2.4梯度提升決策樹(Gradient Boost Decision Tree).......35
2.2.5 樸素貝葉斯(Naive Bayes' Theorem).........36
第三章 所提出機器學習之預測方法..............39
3.1資料前處理.............39
3.2實驗流程..............44
3.3評估與驗證模型的指標....47
第四章 實驗結果.............51
4.1多項驗證指標...........51
4.2特徵重要性.............54
4.3ROC、PR與其AUC指標.....59
第五章 結論.................61
參考文獻...................63

[1]A. Oreziak, E. Piatkowska-Janko, Z. Lewandowski and G Opolski, “Prediction of the supraventricular arrhythmias in hypertensive patients with different forms of the left ventricular geometry,” Computers in Cardiology, Valencia, Spain, May 2008.
[2]J. R. Sowers, M. Epstein and E. D. Frohilich, “Diabetes, hypertension, and cardiovascular disease an update,” Hypertension, vol. 37, no. 4, pp.1053-1059, 2001.
[3]A. J. Atkinson, S. R. Kharche, M. G. Bateman, P. A. Iaizzo and H. Dobrzynski, “3D anatomical reconstruction of human cardiac conduction system and simulation of bundle branch block after TAVI procedure,” IEEE Engineering in Medicine and Biology Society, Orlando, FL, USA, October 2016.
[4]Y. Liu and Pinpin Tang, “The prospect for the application of the surgical navigation system based on artificial intelligence and augmented reality,” IEEE International Conference on Artificial Intelligence and Virtual Reality, Taichung, Taiwan, January 2019.
[5]C. J. Huang, M. C. Liu, S. S. Chu and C. L. Cheng, “Application of machine learning techniques to web-based intelligent learning diagnosis system,” International Conference on Hybrid Intelligent Systems, Kitakyushu, Japan, April 2005.
[6]T. D. Bui, D. K. Nguyen and T. D. Ngo, “Supervising and unsupervised neural network,” Intelligent Information and Database Systems, Dong Hoi, Vietnam, July 2009.
[7]Z. Wu, N. M. Khan, L. Gao and Ling Guan, “Deep reinforcement learning with parameterized action space for object detection,” IEEE International Symposium on Multimedia, Taichung, Taiwan, January 2019.
[8]C. Tobon-Gomez , C. Butakoff , P. Yushkevich , M. Huguet and A.F. Frangi,“3D Mesh based wall thickness measurement: identification of left ventricular hypertrophy phenotypes,” Annual International Conference of the IEEE Engineering in Medicine and Biology, Buenos Aires, Argentina, November 2010.
[9]A. Muankid and M. Ketcham, “The limb leads ECG signal analysis in inferior myocardial infarction patients by rule base,” International Joint Symposium on Artificial Intelligence and Natural Language Processing, Pattaya, Thailand, April 2019.
[10]M. W. Gifari, H. Aakaria and R. Mengko, “Design of ECG homecar:12-lead ECG acquisition using single channel ECG device developed on AD8232 analog front end,” International Conference on Electrical Engineering and Informatics, Denpasar, Indonesia, August. 2015.
[11]P. Kaewfoongrungsi and D. Hormdee, “The comparison between linear regression derivings of 12-lead ECG signals from 5-lead system and EASI-lead system,” International Symposium on Communications and Information Technologies, Cairns, QLD, Australia, January 2018.
[12]C. Yao, J. M. Simpson, T. Schaeffter and G. P. Penney, “Spatial compounding of large numbers of multi-view 3D echocardiography images using feature consistency,” IEEE International Symposium on Biomedical Imaging: From Nano to Macro, Rotterdam, Netherlands, June 2010.
[13]P. Joos, H. Liebgott, F. Varray, L. Petrusca, D. Garcia, D. Vray and B. Nicolas, “High-frame-rate 3-D echocardiography based on motion compensation: an in vitro evaluation,” IEEE International Ultrasonics Symposium, Washington DC, USA, November 2017.
[14]M. Foppa, B. B. Duncan and L. EP Rohde, “Echocardiography-based left ventricular mass estimation. How should we define hypertrophy?,” Cardiovascular Ultrasound, vol. 3, no. 17, pp.3-5, June 2005.
[15]J. K. Park, J. H. Shin, S. H. Kim, Y. H. Lim, K. S. Kim , S. G. Kim. J. H. Kim, H. G. Lim and J. Shin, “A comparison of Cornell and Sokolow-Lyon electrocardiographic criteria for left ventricular hypertrophy in Korean patients,” Korean Circulation Journal, vol. 42, no. 9, pp.606-613, April 2004.
[16] F. Y. Su, Y. H. Li, Y. P. Lin, C. J. Lee, C. H. Wang, F. C. Meng, Y. S. Yu, F. Lin, H. T. Wu and G. M. Lin, “A comparison of Cornell and Sokolow-Lyon electrocardiographic criteria for left ventricular hypertrophy in a military male population in Taiwan: the CHIEF study,” Cardiovascular Diagnosis and Therapy, vol. 7, no. 3, pp. 244-251, July 2017.
[17]H. R. Guo, Z. M. Li, “A method of improving generalization ability for neural network based on genetic algorithm,” IEEE International Conference on Intelligent Computing and Intelligent Systems, Xiamen, China, December 2010.
[18]R. Purwahana, S. Suryono and J. E. Suseno, “Cloud-based multinomial logistic regression for analyzing maternal mortality data in postpartum period,” International Conference on Informatics and Computational Sciences, Semarang, Indonesia, January 2019.
[19]D. G. Kleinbaum and M. Klein, Maximum likelihood techniques: an overview, New York, NY: Springer, 2010, pp.103-127.
[20]A. Iliev, N. Kyurkchiev and S. Markov, “On the approximation of the step function by some sigmoid functions,” Mathematics and Computers in Simulation, vol. 133, pp.223-234, March 2017.
[21]Y. LeCun and Y. Bengio, Deep learning, MA, USA: MIT Press Ltd., 2016, pp.227-230.
[22]J. R. Quinlan, “Induction of decision trees,” Machine Learning, vol. 1, no.1, pp.81-106, March 1986.
[23]G. Ke, A. Meng, T. Finley, T. Wang, W. Chen, W. Ma, Q. Ye and T. Y. Liu, “LightGBM: A highly efficient gradient boosting decision tree,” Advances in Neural Information Processing System, pp3146-3154, January 2017.
[24]S. Taheri and M. Mammadov, “Learning the naïve Bayes classifier with optimization models,” International Journal of Applied Mathematics and Computer Science, vol. 23, no. 4, pp.787-795, 2013.
[25]T. Jayalakshmi and Dr. A. Santhakumaran, “Statistical normalization and back propagation for classification,” International Journal of Computer Theory and Engineering, vol. 3, no. 1, pp.1793-820, February 2011.
[26]Y. K. Jain and S. K. Bhandare, “Min max normalization based data perturbation method for privacy protection,” International Journal of Computer & Communication Technology, vol. 2, no. 8, pp. 3146-3154, January 2011.
[27]S. Barua, M. M. Islam, X. Yao and K. Murase, ”MWMOTE-Majority weighted minority oversampling technique for imbalanced data set learning,” IEEE Transactions on Knowledge and Data Engineering, vol. 26, no. 2, pp. 405-425, November 2012.
[28]N. V. Chawla, K. W. Bowyer, L. O. Hall, W. P. Kegelmeyer, “SMOTE: synthetic minority over-sampling technique,” Journal of Artificial Intelligence Research, vol. 16, pp. 321-357, June 2002.
[29]S. Visa, B. Ramsay, A. Ralescu and Esther V. D, Knaap, “Confusion matrix-based feature selection,” Midwest Artificial Intelligence and Cognitive Science Conference, vol. 710, pp120-127, April 2011.
[30]E. Keedwell, “An analysis of the area under the roc curve and its use as a metric for comparing clinical scorecards,” IEEE International Conference on Bioinformatics and Biomedicine, Belfast, UK, January 2015.
[31]J. Grau, I. Grosse and J. Keilwagen, “PRROC: computing and visualizing precision-recall and receiver operating characteristic curves in R,” Bioinformatics, vol. 31, no. 15, pp.2595-2597, March 2015.
[32]A. Jain and D. Zongker, “Feature selection: evaluation, application, and small sample performance,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 19, no. 2, pp.153-158, February 1997.
[33]V. Manoliu and C. M. Stanescu, “A study of the cardiovascular function and aerobic capacity in overweight and obese subjects,” International Symposium on Advanced Topics in Electrical Engineering, Bucharest, Romania, May 2019.