Researchers Develop AI System to Integrate Patient and FDA Data on Psychiatric Medication Safety

Researchers created a knowledge-graph-based AI framework that combines patient experiences from Reddit and WebMD with FDA adverse-event records to provide more comprehensive psychiatric medication safety information. The system analyzed 466,525 Reddit posts, 60,782 WebMD reviews, and 20 years of FDA data for nine antidepressants, with an LLM-based entity recognition pipeline achieving 97% accuracy for medications and conditions. Patient-generated data revealed safety signals hundreds of days before FDA reports appeared, suggesting community platforms constitute an independent safety data source.

Why it matters: As patients increasingly rely on online sources for medication information, this research demonstrates how AI can responsibly integrate regulated evidence with patient narratives while maintaining source transparency—critical for psychiatry where misinformation drives non-adherence and nocebo effects.