Сравнительный анализ нейронных сетей в задаче классификации побочных эффектов на уровне сущностей в англоязычных текстах

В данной работе представлено экспериментальное исследование эффективности ряда моделей нейронных сетей для задачи классификации побочных эффектов на уровне сущностей. Задача анализа тональности на уровне аспектных терминов, в которых необходимо определить мнение по отношению к конкретному аспекту, активно исследуется в течении последнего десятилетия. Для решения данной задачи в прошедшие годы было предложено несколько архитектур нейронных сетей. Несмотря на то, что модели, основанные на этих архитектурах, имеют много общего, есть некоторые компоненты, которые отличают их друг от друга. В данной статье была исследована применимость разработанных для аспектно ориентированного анализа тональности нейросетевых моделей для классификации побочных эффектов. Для оценки эффективности данных методов были проведены обширные эксперименты на различных англоязычных текстах биомедицинской тематики, включающих в себя записи клинических карточек, научную литературу и данные из социальных сетей. Также мы сравнили предлагаемую модель с одной из наилучших на данный момент моделей, основанной на методе опорных векторов и большом наборе признаков.

1. Murff HJ, Patel VL, Hripcsak G, Bates DW. Detecting adverse events for patient safety research: a review of current methodologies. Journal of Biomedical Informatics, vol. 36, issue ½, 2003, pp. 131-143.

2. Sarker A, Ginn R, Nikfarjam A, O’Connor K, Smith K, Jayaraman S et al. Utilizing social media data for pharmacovigilance: A review. Journal of Biomedical Informatics, vol. 54, 2015, pp. 202-212.

3. Lardon J, Abdellaoui R, Bellet F, Asfari H, Souvignet J, Texier N et al. Adverse Drug Reaction Identification and Extraction in Social Media: A Scoping Review. Journal of Medical Internet Research, vol 17, no 7, 2015.

4. Harpaz R, Callahan A, Tamang S, Low Y, Odgers D, Finlayson S, et al. Text mining for adverse drug events: the promise, challenges, and state of the art. Drug Safety, vol. 37, 2014, pp. 777-790.

5. Harpaz R, DuMouchel W, Shah NH, Madigan D, Ryan P, Friedman C. Novel data-mining methodologies for adverse drug event discovery and analysis. Clinical Pharmacology & Therapeutics, vol. 91, no. 6, 2012, pp. 1010-1021.

6. Tang D, Qin B, Feng X, Liu T. Effective LSTMs for Target-Dependent Sentiment Classification [Internet]. arXiv [cs.CL], 2015. Available at: http://arxiv.org/abs/1512.01100, accessed 15.11.2008

7. Ma D, Li S, Zhang X, Wang H. Interactive attention networks for aspect-level sentiment classification. arXiv preprint arXiv:1709 00893, 2017.

8. Tang D, Qin B, Liu T. Aspect level sentiment classification with deep memory network. arXiv preprint arXiv:1605 08900, 2016;

9. Chen P, Sun Z, Bing L, Yang W. Recurrent attention network on memory for aspect sentiment analysis. In Proc. of the 2017 Conference on Empirical Methods in Natural Language Processing. 2017, pp. 452-461.

10. Benton A, Ungar L, Hill S, Hennessy S, Mao J, Chung A, et al. Identifying potential adverse effects using the web: a new approach to medical hypothesis generation. Journal of Biomedical Informatics, vol. 44, 2011, pp. 989-996.

11. Yang CC, Yang H, Jiang L, Zhang M. Social Media Mining for Drug Safety Signal Detection. In Proc. of the 2012 International Workshop on Smart Health and Wellbeing, 2012. pp. 33-40.

12. Liu X, Chen H. AZDrugMiner: An Information Extraction System for Mining Patient-Reported Adverse Drug Events in Online Patient Forums / Lecture Notes in Computer Science, vol. 8040, 2013. pp. 134-150.

13. Yeleswarapu S, Rao A, Joseph T, Saipradeep VG, Srinivasan R. A pipeline to extract drug-adverse event pairs from multiple data sources. BMC Medical Informatics and Decision Making, vol. 14, no. 13, 2014.

14. Freifeld CC, Brownstein JS, Menone CM, Bao W, Filice R, Kass-Hout T, et al. Digital drug safety surveillance: monitoring pharmaceutical products in twitter. Drug Safety, vol. 37, 2014, pp. 343-350.

15. O’Connor K, Pimpalkhute P, Nikfarjam A, Ginn R, Smith KL, Gonzalez G. Pharmacovigilance on twitter? Mining tweets for adverse drug reactions. In Proc. of the AMIA Annual Symposium, 2014, pp. 924-933.

16. Nikfarjam A, Gonzalez GH. Pattern mining for extraction of mentions of Adverse Drug Reactions from user comments. In Proc. of the AMIA Annual Symposium, 2011, pp. 1019-1026.

17. Na J-C, Kyaing WYM, Khoo CSG, Foo S, Chang Y-K, Theng Y-L. Sentiment Classification of Drug Reviews Using a Rule-Based Linguistic Approach. Lecture Notes in Computer Science, vol. 7634, 2012. pp. 189-198.

18. Yun Niu et al. Analysis of polarity information in medical text. In Proc. of the AMIA Annual Symposium, 2005, pp. 570-574.

19. Leaman R. et al. Towards internet-age pharmacovigilance: extracting adverse drug reactions from user posts to health-related social networks. In Proc. of the 2010 workshop on biomedical natural language processing, 2010, pp. 117-125.

20. Yun Niu, Xiaodan Zhu et al. Predicting adverse drug events from personal health messages. In Proc. of the AMIA Annual Symposium, 2011, pp. 217-226.

21. Bian J., Topaloglu U., Yu F. Towards large-scale twitter mining for drug-related adverse events. In. Proc. of the 2012 International workshop on smart health and wellbeing, 2012, pp. 25-32.

22. Yang M., Wang X., Kiang M. Y. Identification of Consumer Adverse Drug Reaction Messages on Social Media. In Proc. of the Pacific Asia Conference on Information Systems, 2013.

23. Sarker A., Gonzalez G. Portable automatic text classification for adverse drug reaction detection via multi-corpus training. Journal of biomedical informatics, vol. 53, 2015, pp. 196-207.

24. Aramaki E. et al. Extraction of adverse drug effects from clinical records. Studies in Health Technology and Informatics, vol. 160, №. Pt 1, 2010, pp. 739-743.

25. Rastegar-Mojarad M., Elayavilli R.K., Yu Y., Liu H. Detecting signals in noisy data-can ensemble classifiers help identify adverse drug reaction in tweets. In Proc. of the Social Media Mining Shared Task Workshop at the Pacific Symposium on Biocomputing, 2016.

26. Sarker A, Nikfarjam A, Gonzalez G. Social Media Mining Shared Task Workshop. In Proc. of the Pacific Symposium on Biocomputing, 2016, pp. 581-592.

27. Sarker A, Gonzalez-Hernandez G. Overview of the Second Social Media Mining for Health (SMM4H) Shared Tasks at AMIA 2017. In Proc. of the 2nd Social Media Mining for Health Research and Applications Workshop, 2017, pp. 43-48.

28. Kiritchenko S, Mohammad SM, Morin J, de Bruijn B. NRC-Canada at SMM4H Shared Task: Classifying Tweets Mentioning Adverse Drug Reactions and Medication Intake. arXiv preprint arXiv:1805 04558. 2018.

29. Friedrichs J, Mahata D, Gupta S. InfyNLP at SMM4H Task 2: Stacked Ensemble of Shallow Convolutional Neural Networks for Identifying Personal Medication Intake from Twitter. arXiv preprint arXiv:1803 07718. 2018.

30. Huynh T, He Y, Willis A, Rüger S. Adverse drug reaction classification with deep neural networks. In Proc. of the 26th International Conference on Computational Linguistics: Technical Papers, 2016, pp. 877-887.

31. Gurulingappa H., Rajput A.M. et al. Development of a benchmark corpus to support the automatic extraction of drug-related adverse effects from medical case reports. Journal of Biomedical Informatics, vol. 45, 2012, pp. 885-892.

32. Serrano-Guerrero J., Olivas J.A. et al. Sentiment analysis: A review and comparative analysis of web services. Information Sciences, vol. 311, 2015, pp. 18-38

33. Rusnachenko N., Loukachevitch N. Using convolutional neural networks for sentiment attitude extraction from analytical texts. In Proc. of the Third Workshop on Computational linguistics and language science (to be published in CEUR Workshop Proceedings), 2018

34. Ivanov V., Tutubalina E., Mingazov N., Alimova I. Extracting aspects, sentiment and categories of aspects in user reviews about restaurants and cars. Computational Linguistics and Intellectual Technologies. Papers from the Annual International Conference “Dialogue”, issue 14, vol. 2, 2015, pp. 22-34

35. Solovyev V., Ivanov V. Dictionary-based problem phrase extraction from user reviews. Lecture Notes in Computer Science, vol. 8655, 2014, pp. 225-232.

36. Zhang L., Wang S., Liu, B. Deep learning for sentiment analysis. A survey. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, vol. 8, issue 4, 2018.

37. Alimova I., Tutubalina E. Automated detection of adverse drug reactions from social media posts with machine learning. Lecture Notes in Computer, vol. 10716, 2017, pp. 3-15.

38. Miftahutdinov Z.S., Tutubalina E.V., Tropsha A.E. Identifying disease-related expressions in reviews using conditional random fields. Computational Linguistics and Intellectual Technologies: Papers from the Annual conference “Dialogue”, issue 16, vol. 1, 2017, pp 155-166

39. Korkontzelos I., Nikfarjam A. et al. Analysis of the effect of sentiment analysis on extracting adverse drug reactions from tweets and forum posts. Journal of biomedical informatics, vol. 62, 2016, pp. 148-158.

40. Dai H.-J., Touray M., Jonnagaddala J., Syed-Abdul S. Feature engineering for recognizing adverse drug reactions from twitter posts. Information, vol. 7, no. 27, 2016.

41. Karimi, S. Metke-Jimenez,A., Kemp M., Wang C.: Cadec. A corpus of adverse drug event annotations. Journal of biomedical informatics, vol. 55, 2015, pp. 73-81.

42. Nikfarjam A., Sarker A. et al. Pharmacovigilance from social media: mining adverse drug reaction mentions using sequence labeling with word embedding cluster features. Journal of the American Medical Informatics Association, vol. 22, no. 3, 2015, pp. 671-681

43. Nlp challenges for detecting medication and adverse drug events from electronic health records (made1.0) (2018). University of Massachusetts Lowell, Worcester, Amhers. Available at: https://bio-nlp.org/index.php/projects/39-nlp-challenges, accessed 15.11.2008.

44. Alvaro N., Miyao Y., Collier N. Twimed: Twitter and pubmed comparable corpus of drugs, diseases, symptoms, and their relations. JMIR public health and surveillance, vol. 3, no. 2, 2017.

45. Wilson T., Wiebe J., Hoffmann P. Recognizing contextual polarity in phrase-level sentiment analysis. In Proc. of the Conference on Human Language Technology and Empirical Methods in Natural Language Processing, 2005, pp. 347-

46. Baccianella S., Esuli A., Sebastiani F. Sentiwordnet 3.0: an enhanced lexical resource for sentiment analysis and opinion mining. In Proceedings of the Seventh conference on International Language Resources and Evaluation, 2010, pp. 2200-2204 (2010)

47. Hu M., Liu B. Mining and summarizing customer reviews. In Proc, of the Tenth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2004, pp. 168-177.