Multi-Agent AI Tracks Antidepressant Side Effects from Reddit, WebMD, and FDA

Patients increasingly turn to online communities for medication information, but psychiatric drug safety knowledge is split between authoritative yet abstract regulatory adverse-event records and experience-near but unvalidated patient narratives. A team of researchers developed a provenance-aware, knowledge-graph-based multi-agent framework to integrate these disparate sources without conflating evidence and anecdote. The system ingests 466,525 Reddit posts, 60,782 WebMD reviews, and twenty years of U.S. FDA Adverse Event Reporting System records for nine antidepressants. Using a large-language-model entity-recognition pipeline benchmarked against physician annotations, it achieved highest F1 scores of 0.969 for medications and 0.973 for conditions. The two community platforms were far more concordant with each other (overlap up to Jaccard similarity of 0.905) than with regulatory reports, indicating that patient-generated data form a partly independent safety signal.

The framework's practical impact is significant: for sertraline, many adverse events appeared in community sources hundreds of days before the corresponding FDA date. A Neo4j knowledge graph grounded in ATC-N, ICD-10, and MedDRA vocabularies preserves provenance, keeping every claim traceable and regulatory facts distinct from patient experience. These results establish source-aware integration as a route to more auditable psychiatric medication information. The authors note that usefulness and patient benefit will be tested prospectively, but the approach already demonstrates how AI can bridge the gap between formal safety monitoring and real-world patient experiences, potentially reducing fear, nocebo responses, and non-adherence through better contextualized information.