Pharmacovigilance from social media: mining adverse drug reaction mentions using sequence labeling with word embedding cluster features

Abstract

Objective Social media is becoming increasingly popular as a platform for sharing personal health-related information. This information can be utilized for public health monitoring tasks, particularly for pharmacovigilance, via the use of natural language processing (NLP) techniques. However, the language in social media is highly informal, and user-expressed medical concepts are often nontechnical, descriptive, and challenging to extract. There has been limited progress in addressing these challenges, and thus far, advanced machine learning-based NLP techniques have been underutilized. Our objective is to design a machine learning-based approach to extract mentions of adverse drug reactions (ADRs) from highly informal text in social media.Methods We introduce ADRMine, a machine learning-based concept extraction system that uses conditional random fields (CRFs). ADRMine utilizes a variety of features, including a novel feature for modeling words’ semantic similarities. The similarities are modeled by clustering words based on unsupervised, pretrained word representation vectors (embeddings) generated from unlabeled user posts in social media using a deep learning technique.Results ADRMine outperforms several strong baseline systems in the ADR extraction task by achieving an F-measure of 0.82. Feature analysis demonstrates that the proposed word cluster features significantly improve extraction performance.Conclusion It is possible to extract complex medical concepts, with relatively high performance, from informal, user-generated content. Our approach is particularly scalable, suitable for social media mining, as it relies on large volumes of unlabeled data, thus diminishing the need for large, annotated training data sets.

Natural Language Analysis of Online Health Forums

“…Prior research in finding useful health related information from social media, mainly focused on named entity extraction (NER) tasks such as discovering ADRs in relation to a drug or treatment as reported in [8]. For example, Nikfarjam et al [12] built a system, called ADRMine, using CRFs for recognising ADR mentions from social media. Features such as contextual, lexical and semantic parts-of-speech (POS) tags features were added to an existing CRF classifier.…”

“…For example, Gupta et al [7] attempted to extract drugs and treatments (DTs), and symptoms and conditions (SCs) terms present in the forum text. On the other hand, Nikfarajam et al [12] and Sampathkumar et al [16] extracted adverse drug reactions (ADRs) from Twitter and the DailyStrength forum [4]. Their research shows that social media is a valuable source for finding information related to drugs, symptoms and side effects.…”

A Survey of Current Work in Medical Text Mining---Data Source Perspective

“…Nikfarjam A, Sarker A&O'Connor K etc. (2015) attempt to develop a machine learning-based approach to extract mentions of adverse drug reactions (ADRs) from highly informal and descriptive text in social media [5]. They introduce a machine learning-based concept extraction system, namely ADR Mine, to model words' semantic similarities by cluster approach.…”

Detecting Personal Medication Intake in Twitter: An Annotated Corpus and Baseline Classification System

“…A large subset of the public health-related research using social media data, including our prior work in the domain, focuses on mining information (e.g., adverse drug reactions, medication abuse, and user sentiment) from posts mentioning medications (Korkontzelos et al, 2016;Hanson et al, 2013b;Nikfarjam et al, 2015). Typically, these and similar studies focus on information at the population level, but processing and deriving information from individual user posts poses significant challenges from the natural language processing (NLP) perspective.…”