Social media are a robust source of health-related data and may serve as complementary sources of adverse drug reaction (ADR) information from the patient perspective.

Analyses of social media posts allow for open, scientific investigation of ADRs that may be either not reported or underreported in spontaneous reporting systems and primary literature, thus contributing to a more complete safety profile.

Challenges prevent current natural language processing methods to automatically map all consumer ADR expressions to standard forms. Improvements to automatic text processing approaches should make the methods presented here scalable by reducing the annotation burden.