Abstract
Drug side effects are a major global health concern, necessitating advancedmethods for their accurate detection and analysis. While Large Language Models(LLMs) offer promising conversational interfaces, their inherent limitations,including reliance on black-box training data, susceptibility tohallucinations, and lack of domain-specific knowledge, hinder their reliabilityin specialized fields like pharmacovigilance. To address this gap, we proposetwo architectures: Retrieval-Augmented Generation (RAG) and GraphRAG, whichintegrate comprehensive drug side effect knowledge into a Llama 3 8B languagemodel. Through extensive evaluations on 19,520 drug side effect associations(covering 976 drugs and 3,851 side effect terms), our results demonstrate thatGraphRAG achieves near-perfect accuracy in drug side effect retrieval. Thisframework offers a highly accurate and scalable solution, signifying asignificant advancement in leveraging LLMs for critical pharmacovigilanceapplications.