Pharmacovigilance from social media - is it possible?

By William Chin, PhD, Project Manager, EUCRAF

It seems like it is possible to extract complex medical concepts from social media posts. A research paper entitled "Pharmacovigilance from social media: mining adverse drug reaction mentions using sequence labeling with word embedding cluster features" was published in the March 2015 issue of the Journal of the American Medical Informatics Association. An increasing trend is reported where the public is sharing personal health-related information on social media. Nikfarjam A et al. explored the idea of designing an algorithm capable of extracting information on adverse drug reactions (ADR) from informal text in social media. The objective was to use this information for pharmacovigilance. In this study, user posts about drugs were collected from a health social networking site, DailyStrength (DS), and a generic social network, Twitter. The authors demonstrated that it was possible to accurately identify most of the ADR mentions, including the consumer expressions that are not observed in the training data or in the standard ADR lexicons. The extraction performance for DS is much higher than Twitter because the availability of large numbers of annotated data in the health-focused site. Future work will focus on improvising the current features and extraction techniques.

Sources

1. J Am Med Inform Assoc. 2015 Mar 9. Pharmacovigilance from social media: mining adverse drug reaction mentions using sequence labeling with word embedding cluster features. Nikfarjam A, Sarker A, O'Connor K, Ginn R, Gonzalez G.