Adverse Drug Reaction Detection in Social Media by Deep Learning Methods

Rezaei, Zahra; Eslami, Behnaz; Chavoshinejad, Ramyar; Totonchi, Mehdi

doi:10.22074/cellj.2020.6615

Adverse Drug Reaction Detection in Social Media by Deep Learning Methods

Document Type : Original Article

Authors

¹ .Department of Computer and Electrical Engineering, University of Kashan, Kashan, Iran

² .Department of Computer Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran

³ .Mabna Veterinary Lab, Karaj, Alborz, Iran

⁴ 4.Department of Genetics, Reproductive Biomedicine Research Center, Royan Institute for Reproductive Biomedicine, ACECR, Tehran, Iran;5.Department of Stem Cells and Developmental Biology, Cell Science Research Center, Royan Instit

10.22074/cellj.2020.6615

Abstract

Objective
Health-related studies have been recently at the heart attention of the media. Social media, such as Twitter, has become a valuable online tool to describe the early detection of various adverse drug reactions (ADRs). Different medications have adverse effects on various cells and tissues, sometimes more than one cell population would be adversely affected. These types of side effect are occasionally associated with the direct or indirect influence of prescribed drugs but do not have general unfavorable mutagenic consequences on patients. This study aimed to demonstrate a quick and accurate method to collect and classify information based on the distribution of approved data on Twitter.
Materials and Methods
In this classification method, we selected "ask a patient" dataset and combination of Twitter "Ask a Patient" datasets that comprised of 6,623, 26,934, and 11,623 reviews. We used deep learning methods with the word2vec to classify ADR comments posted by the users and present an architecture by HAN, FastText, and CNN.
Results
Natural language processing (NLP) deep learning is able to address more advanced peculiarity in learning information compared to other types of machine learning. Moreover, the current study highlighted the advantage of incorporating various semantic features, including topics and concepts.
Conclusion
Our approach predicts drug safety with the accuracy of 93% (the combination of Twitter and "Ask a Patient" datasets) in a binary manner. Despite the apparent benefit of various conventional classifiers, deep learning- based text classification methods seem to be precise and influential tools to detect ADR.

Keywords