Tag
This paper presents a provenance-aware, knowledge-graph-based multi-agent framework that integrates patient narratives from Reddit and WebMD with FDA adverse event reports for nine antidepressants, using an LLM entity-recognition pipeline to achieve high accuracy and enabling traceable safety information for psychiatric medications.