Artificial Intelligence in Pharmacovigilance: Using AI to help GSK detect ICSRs in scientific literature

GlaxoSmithKline needed to review thousands of abstracts daily to identify Individual Case Safety Reports (ICSRs) as part of its Market Authorisation Holder obligations, using Clarivate's Dialog platform for search and Drug Safety Triager for manual review — a growing burden given that roughly 6,000 peer-reviewed articles publish daily and only about 5% of search results contain a valid ICSR. GSK and Clarivate ran a proof of concept of DialogML, a natural-language-processing relevancy ranking engine, using GSK's previously annotated literature abstracts as test/training data. Results: 99% of ICSRs were found within the top 33% of ranked abstracts and 100% within the top 65%, meaning a reviewer could find 85% of valid ICSRs within the first hour versus 12.5% without DialogML; average time each reference spent in the review queue dropped from 4.5 hours to 1.2 hours. Following the PoC, GSK signed up to use DialogML as part of its medical literature monitoring process by upgrading to the latest version of Drug Safety Triager.

Clarivate owns Cortellis, the curated competitive-intelligence and pipeline-tracking platform standard in pharma portfolio-management teams.

This sits in the Data Providers stage of the AI use-case lifecycle: Named examples from the real-world-data, claims and competitive-intelligence providers pharma AI systems are trained on, benchmarked against, or run inside of.