臺灣博碩士論文加值系統

藥物不良反應在全世界一直以來都是一個嚴重的議題，不同於藥物副作用，藥物不良反應專指患者在使用某種藥物產生對於非預期且對人體造成傷害的作用，可以造成感冒症狀、皮膚紅疹、甚至是死亡，對於免疫力下降且同時採用不同治療的高齡人口更為致命。所有的藥物都存在藥物不良的風險，其症狀會因病患個體的差異產生不同反應，本研究致力於建立一個藥物不良感應的判別系統，透過出院病摘之資料，辨別出是否存有藥物不良反應，期望可以為藥物不良反應相關領域做出貢獻。
為了偵測出臨床的出院病例中的藥物不良反應，我們提出了一個結合詞嵌入將文字轉化為能以深度學習處理之維度，透過卷積神經網路及深度支援向量資料描述結合形成之分類器混合模型，利用詞嵌入將資料降為致特徵維度並將其數位化，使語意上越相似的字詞越靠近彼此，再先經由自編碼作為預訓練，最後接上卷積神經網路與深度支援向量資料描述作為分類之依據，藉由深度支援向量資料描述可以辨別出該病歷摘要是否存有藥物不良反應。
本研究之結果與隱含狄利克雷分布連結單類別支援向量機進行比較，於AUROC以及AUPRC兩個面向都有較優越的結果。未來，專家可以透過本研究之分類器結果加以驗證，以提升藥物不良反應通報的精確性。

Adverse Drug Reactions (ADRs) have always been a serious issue in the world. Unlike drug side effects, ADRs are specific to patients who use certain drugs to produce unintended and harmful effects on the human body. ADRs can cause cold symptoms, red skin, uncomfortable, illness, or even death. Especially it is fatal to the aged people due to lower immunity and polypharmacy. Because all drugs have potential of ADRs, this study aims to establish a discriminating system for predicting ADRs with clinical data. The reaction is expected to contribute to the field of ADRs.
To detect ADRs in discharge summaries, we propose a hybrid model by combining with word2vec and Deep-SVDD to convert words into dimensions that can be processed in deep learning, and to form a classifier through Convolutional Neural Network (CNN) and Deep Support Vector Data Description (Deep-SVDD). Word2vec provides the ability to embed the words into vector space which is a suitable input for Convolutional neural network(CNN). Deep-SVDD, which is originally transformed from Support Vector Data Description(SVDD) but deepen it with CNN, is applied to detect abnormal information differ from the training data. The proposed model can distinguish whether ADRs occurred by the discharge summaries. The result of our research is much superior than the result of combination of latent Dirichlet allocation (LDA) and One-Class Support Vector Machine (OC-SVM) on both AUROC and AUPRC evaluation. In the future, experts can verify the results of the classifiers in this study to improve the accuracy of ADRs notifications.

摘要 I
Abstract II
Table of Contents IV
List of Tables VII
List of Figures VIII
Chapter 1 Introduction 1
1.1 Background and motivation 1
1.2 Research objectives 2
1.3 Research process 3
Chapter 2 Literature Review 4
2.1 Adverse Drug Reactions (ADRs) 4
2.2 Word representation 5
2.2.1 Dictionary lookup 5
2.2.2 One-hot encoding 6
2.2.3 Distributional representation 7
2.3 Convolutional Neural Network 9
2.4 One-Class classification 11
2.4.1 One-class SVM 11
2.4.2 Support vector data description (SVDD) 12
2.4.3 Deep one-class classification 13
Chapter 3 Research Model 16
3.1 Problem Definition 17
3.2 Data Preprocessing 17
3.2.1 Segmentation ＆ P-O-S tagging 18
3.2.2 Stop-words list 18
3.2.3 Lemmatization 19
3.2.4 Constructing lexicon 19
3.3 Word2Vec 19
3.4 Classifier Construction 21
3.4.1 Optimization 22
3.4.2 Model algorithm 23
Chapter 4 Experiment and Analysis 25
4.1 Experiment Implement 25
4.1.1 Experiment Architecture 25
4.1.2 Data Set Description 26
4.1.3 Word embedding 28
4.1.4 Prediction model construct 29
4.2 Experiment results and Analysis 30
Chapter 5 Conclusion and Future work 36
5.1 Conclusion 36
5.2 Managerial implication 37
5.3 Future work and limitation 37
Reference 39

Bate, A, ＆ Evans, S. J. W. (2009). Quantitative signal detection using spontaneous ADR reporting. Pharmacoepidemiology drug safety, 18(6), 427-436.
Beijer, H. J. M. , ＆ De Blaey, C. J. (2002). Hospitalisations caused by adverse drug reactions (ADR): a meta-analysis of observational studies. Pharmacy World Science, 24(2), 46-54.
British Medical Association. (2006). Reporting adverse drug reactions: a guide for healthcare professionals. In Reporting adverse drug reactions: a guide for healthcare professionals: BMA.
Chalapathy, R., Menon, A. K., ＆ Chawla, S. (2018). Anomaly Detection using One-Class Neural Networks. arXiv preprint arXiv:1802.06360.
Chandola, V., Banerjee, A., ＆ Kumar, V. (2009). Anomaly detection: A survey. ACM computing surveys, 41(3), 15.
Chiu, B., Crichton, G., Korhonen, A., ＆ Pyysalo, S. (2016). How to train good word embeddings for biomedical NLP. Paper presented at the Proceedings of the 15th workshop on biomedical natural language processing.
Fawcett, T. (2006). An introduction to ROC analysis. Pattern recognition letters, 27(8), 861-874.
Hajjar, E. R., Cafiero, A. C., ＆ Hanlon, J. T. (2007). Polypharmacy in elderly patients. The American journal of geriatric pharmacotherapy, 5(4), 345-351.
Han, J., Pei, J., ＆ Kamber, M. (2011). Data mining: concepts and techniques: Elsevier.
Kalchbrenner, N., Grefenstette, E., ＆ Blunsom, P. (2014). A convolutional neural network for modelling sentences. arXiv preprint arXiv:1404.2188.
Khan, S. S., ＆ Madden, M. G. (2009). A survey of recent trends in one class classification. Paper presented at the Irish conference on artificial intelligence and cognitive science.
Khan, S. S., ＆ Madden, M. G. (2014). One-class classification: taxonomy of study and review of techniques. The Knowledge Engineering Review, 29(3), 345-374.
Kingma, D. P., ＆ Ba, J.(2014). Adam: A method for stochastic optimization.
Lawrence, S., Giles, C. L., Tsoi, A. C., ＆ Back, A. D. (1997). Face recognition: A convolutional neural-network approach. IEEE transactions on neural networks, 8(1), 98-113.
Liou, C. Y., Cheng, W. C., Liou, J. W., ＆ Liou, D. R. (2014). Autoencoder for words. Neurocomputing, 139, 84-96.
Mikolov, T., Chen, K., Corrado, G., ＆ Dean, J. (2013). Efficient estimation of word representations in vector space. arXiv preprint arXiv:.1301.3781.
Mikolov, T., Sutskever, I., Chen, K., Corrado, G. S., ＆ Dean, J. (2013). Distributed representations of words and phrases and their compositionality. Paper presented at the Advances in neural information processing systems.
Mohammad, S., ＆ Hirst, G. (2006). Distributional measures of concept-distance: A task-oriented evaluation. Paper presented at the Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing.
Ng, A. (2011). Sparse autoencoder. CS294A Lecture notes. https://web.stanford.edu/class/cs294a/sparseAutoencoder_2011new.pdf
Ruff, L., Görnitz, N., Deecke, L., Siddiqui, S. A., Vandermeulen, R., Binder, A., Kloft, M. (2018). Deep one-class classification. In International Conference on Machine Learning, pages 4390–4399, 2018a.
Sahlgren, M. (2008). The distributional hypothesis. Italian Journal of Disability Studies, 20, 33-53.
Schölkopf, B., Platt, J. C., Shawe-Taylor, J., Smola, A. J., ＆ Williamson, R. C. (2001). Estimating the support of a high-dimensional distribution. Journal of Neural computation, 13(7), 1443-1471.
Schatz, S., ＆ Weber, R. (2015). Adverse drug reactions. Pharmacy Practice.
Schlkopf, B., Williamson, R., Smola, A., Shawe Taylor, J., ＆ Platt, J. (1999). Support vector method for novelty detection. Paper presented at the Proceedings of the 12th International Conference on Neural Information Processing Systems, Denver, CO.
Schnabel, T., Labutov, I., Mimno, D., ＆ Joachims, T. (2015). Evaluation methods for unsupervised word embeddings. Paper presented at the Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing.
Shao, Y., Hardmeier, C., Tiedemann, J., ＆ Nivre, J. (2017). Character-based joint segmentation and POS tagging for Chinese using bidirectional RNN-CRF. arXiv preprint arXiv:.1704.01314.
Sharir, O., Tamari, R., Cohen, N. , ＆ Shashua, A. (2016). Tensorial mixture models. Arxiv preprint arxiv:1610.04167.
Tang, D., Wei, F., Yang, N., Zhou, M., Liu, T., ＆ Qin, B. (2014). Learning sentiment-specific word embedding for twitter sentiment classification. Paper presented at the Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).
Tax, D.M.J., ＆ Duin, R.P.W. (2001). Uniform object generation for optimizing one-class classifiers. Journal of machine learning research, 2(Dec), 155-173.
Tax, D.M.J., ＆ Duin, R.PW. (2004). Support vector data description. Machine learning, 54(1), 45-66.
Wang, Q.H., Lopes, L. S., ＆ Tax, D.M.J. (2004). Visual object recognition through one-class learning. Paper presented at the International Conference Image Analysis and Recognition.
Wang, Y., Liu, S., Afzal, N., Rastegar-Mojarad, M., Wang, L.W., Shen, F., Liu, H.F. (2018). A comparison of word embeddings for the biomedical natural language processing. Journal of biomedical informatics, 87, 12-20.
World Health Organization. (2002). Safety of medicines: a guide to detecting and reporting adverse drug reactions: why health professionals need to take action. Retrieved from https://apps.who.int/iris/handle/10665/67378
Yang, C. C., Yang, H.D., Jiang, L., ＆ Zhang, M. (2012). Social media mining for drug safety signal detection. Paper presented at the Proceedings of the 2012 international workshop on Smart health and wellbeing.
Yih, W. T., He, X.D., ＆ Meek, C. (2014). Semantic parsing for single-relation question answering. Paper presented at the The 52nd Annual Meeting of the Association for Computational Linguistics Baltimore.
Zeiler, M. D., ＆ Fergus, R. (2013). Stochastic pooling for regularization of deep convolutional neural networks. arXiv preprint arXiv:.1301.3557.