MEDICAL SENTIMENT ANALYSIS BASED ON SOFT VOTING ENSEMBLE ALGORITHM

Abstract

Digital information is continuously generated from various sources such as social media, user reviews for services. The processing of this written information to extract user opinions is critical for developing customer satisfaction. In particular, medical services may be improved with customer feedbacks if the user opinions or sentiments are inferred from user reviews. There is an ongoing effort to develop automated software systems to evaluate these customer reviews. Machine Learning (ML) algorithms combined with Natural Language Processing (NLP) techniques are used to assess customer feedbacks. There are many studies related to English language in the literature to evaluate sentiments of user reviews. However, Turkish language needs research and it has abundant search opportunities in terms of sentiment classification. This work develops a soft voting ensemble (SVE) algorithm that combines predictions of Logistic Regression (LR), Random Forest (RF) and Decision Tree (DT) to analyze a newly collected medical review data. The accuracies of sentiment classifications of LR, RF and DT are 90.68%, 89.03% and 85.41%. The sentiment classification accuracy of SVE, combination of three algorithms, is 91.12%. The obtained results are promising for an automated Turkish medical sentiment identification algorithm.

Keywords

• Medical Information,Sentiment Classification,Ensemble Learning,Soft Voting

References

1. Al-Ayyoub, M., Khamaiseh, A.A., Jararweh & Y., Al-Kabi, M. (2019). A comprehensive survey of arabic sentiment analysis, Information Processing and Management: 320-342.
2. Bomaccorso, G. (2018). Machine Learning Algorithms: Popular algorithms for data science and machine learning, Packt Publishing, Birmingham, United Kingdom: 281-282.
3. Coban, O., & Ozel, S. A. (2018). An Empirical Study of the Extreme Learning Machine for Twitter Sentiment Analysis, International Journal of Intelligent Systems and Applications in Engineering: 178-184.
4. Crannell, C.W., Clark, E., Jones, J., James, T., & Moore, J. (2016). A pattern-matched Twitter analysis of US cancer-patient sentiments, Journal of Surgical Research: 536-542.
5. Das, B. & Chakraborty, S. (2018). An Improved Text Sentiment Classification Model Using TF-IDF and Next Word Negation, Computation and Language, arXiv:1806.06407.
6. Dekharghani, R., Yanikoglu, B., Saygin & Y., Oflazer, K. (2016).Sentiment analysis in Turkish at different granularity levels,Natural Language Engineering: 535-559.
7. Ersahin, B., Aktas, O., Kilinc, D. & Ersahin, M. (2019). A hybrid sentiment analysis method for Turkish,, Turkish Journal of Electrical Engineering & Computer Sciences: 1780-1793.
8. Kaewrod, N. & Kietikul, J. (2018). Improving ID3 Algorithm by Ignoring Minor Instances, International Computer Science and Engineering Conference (ICSEC), Chiang Mai, Thailand.

9. Karcioglu, A. A. & Aydin, T. (2019). Sentiment Analysis of Turkish and English Twitter Feeds Using Word2Vec Model, 2019 27th Signal Processing and Communications Applications Conference (SIU), Sivas, Turkey.
10. Korkontzelos, I., Nikfarjam, A., Shardlow, M., Sarker, A., Ananiadou, S. & Gonzalez, G. (2019). Analysis of the effect of sentiment analysis on extracting adverse drug reactions from tweets and forum posts, Journal of Biomedical Informatics: 148-158.
11. Kumar, S., Yadava, M. & Roy, P.P. (2019). Fusion of EEG response and sentiment analysis of products review to predict customer satisfaction, Information Fusion: 41-52.
12. Makinist, S., Hallac, R., Karakus, B. & Aydın, G. (2017). Preparation of Improved Turkish DataSet for Sentiment Analysis in Social Media, ITM Web of Conferences, 01-03 October 2017.
13. Obulesu, O., Mahendra, M. & ThrilokReddy, M. (2018). Machine Learning Techniques and Tools: A Survey, Proceedings of the International Conference on Inventive Research in Computing Applications (ICIRCA 2018).
14. Omari, M., Al-Hajj, M., Hammami, N. & Sabra, A. (2019). Sentiment Classifier: Logistic Regression for Arabic Services’ Reviews in Lebanon, International Conference on Computer & Information Science (ICCIS), Sakaka, Saudi Arabia.
15. Peng, W., Silan, N. & Zhonghua, S. (2019). Random Forest Classification of Rice Planting Area Using Multi-Temporal Polarimetric Radarsat-2 Data, IEEE International Symposium Geoscience and Remote Sensing (IGARSS), Yokohama, Japan. Rajput, A. (2020). Natural Language Processing, Sentiment Analysis, and Clinical Analytics, Innovation in Health Informatics, Cambridge, Massachusetts: 79-97.
16. Rani, R. & Lobiyal, D. K. (2018).Automatic Construction of Generic Stop Words List for Hindi Text, Procedia Computer Science: 362-370.
17. Ribeiro, M.H. & Coelho, L.S. (2020). Ensemble approach based on bagging, boosting and stacking for short-term prediction in agribusiness time series, Applied Soft Computing Journal.
18. Rodrigues, R.G., Dores, R.M., Camilo-Junior, C. & Rosa, T. (2016). SentiHealth-Cancer: A sentiment analysis tool to help detecting mood of patients in online social networks, International Journal of Medical Informatics: 80-95
19. Sagi, O. & Rokach, L. (2018). Ensemble learning: A survey, WIREs Data Mining Knowledge Discovery: 1-18.
20. Shehu, H.A., Tokat, S., Haidar, S. & Uyaver, S. (2019). Sentiment analysis of Turkish Twitter data, AIP Conference Proceedings.
21. Tabassum, N & Tanvir, A. (2016). A theoretical study on classifier ensemble methods and its applications, 2016 3rd International Conference on Computing for Sustainable Global Development (INDIACom), New Delhi, India.
22. Uysal, A. & Gunal, S. (2014). The impact of preprocessing on text classification, Information Processing & Management: 104-112.
23. Wael, E. & Naymat, G. (2017). The Impact of applying Different Preprocessing Steps on Review Spam Detection, Procedia Computer Science: 273-279.
24. Zafra, S.M.,Valdivia, T., González, M. & Lopez, A. (2019). How do we talk about doctors and drugs? Sentiment analysis in forums expressing opinions for medical domain, Artificial Intelligence in Medicine: 50-57.