Azadeh Nikfarjam
Postdoctoral Scholar
BMIR, Stanford University

ABSTRACT:
Adverse drug reactions (ADRs) are a major public health concern and are among the top causes of morbidity and mortality. Many ADRs, are not identified in pivotal clinical trials and initial drug labels. Post-market drug surveillance platforms such as FDA monitoring rely on voluntary, spontaneous reporting and lack temporal advantage over literature. Thus, practitioners on the frontlines of novel treatments, particularly cancer treatments, operate in the dark conducting ADR surveillance. There is a critical need for timely and accurate detection of ADRs in the post-approval period. Internet community health forums provide a robust social platform for individuals to discuss real-time health concerns and may serve as a resource for computational detection of drug safety-related information. We demonstrate the use of a natural language processing pipeline to identify ADR signals from Inspire, one of the leading health forums, to compare user reporting of ADRs with clinical reporting