Extended Trigger Terms for Extracting Adverse Drug Reactions in Social Media Texts

Abstract

Adverse Drug Reaction (ADR) is a disorder caused by taking medications. Studies have addressed extracting ADRs from social networks where users express their opinion regarding a specific medication. Extracting entities mainly depends on specific terms called trigger terms that may occur before or after ADRs. However, these terms should be extended, especially when examining multiple representation of N-gram. This study aims to propose an extension of trigger terms based on the multiple representation of N-gram. Two benchmark datasets are used in the experiments and three classifiers, namely, support vector machine, Naïve Bayes and linear regression, are trained on the proposed extension. Furthermore, two document representations have been utilized including Term Frequency Inverse Document Frequency (TFIDF) and Count Vector (CV). Results show that the proposed extended trigger terms outperform the baseline by achieving 88% and 69% of F1-scores for the first and second datasets, respectively. This finding implies the effectiveness of the proposed extended trigger terms in terms of detecting new ADRs.

Adverse Drug Reaction Detection Using Latent Semantic Analysis

“…1. The first phase is the preparation of annotated drug reviews where the dataset used is from a benchmark dataset by Yates and Goharian (2013) in which Yousef et al (2019) modified some of some structure by adding more meaningful columns of the data. The second phase will contain pre-processing tasks such as tokenization, stop word removal and stemming.…”

“…The study of Yousef et al (2019) tackled the extraction of ADR from social networks where users express their views on a specific medication. Obtaining entities mainly depends on specific terms that may occur before or after ADR, called trigger terms.…”

“…Consequently, the key limitation behind their studies lies in the dependency of using trigger terms where the semantic aspect could be discarded. The novelty of this study is represented by using LSA instead of trigger terms which have been examined by the literature (Ebrahimi et al, 2016;Kiritchenko et al, 2018;Pain et al, 2016;Plachouras et al, 2016;Yousef et al, 2019). LSA is a technique that has been used for identifying the semantics of terms statistically (Al-Sabahi et al, 2018).…”

“…The rise of social networks has contributed toward expanding the textual information dramatically in the last years. Regular users nowadays would have the ability to freely express their minds toward plenty of subjects (Kiritchenko et al, 2018;Yousef et al, 2019). One of these subjects is the product review where a user can evaluate a specific product and describing its advantages and disadvantages based on his/her experience with the product (Liu et al, 2017).…”

“…Most of the studies have utilized machine learning technique classifiers such as SVM and NB. For the feature space, most of the studies have used the trigger terms (Ebrahimi et al, 2016;Kiritchenko et al, 2018;Pain et al, 2016;Plachouras et al, 2016;Yousef et al, 2019). Yet, using trigger terms only would not be sufficient since it needs to be extended periodically when new side effects or new medical-related entities are being discovered.…”