MEDSYNC-AI: A RETRIEVAL-AUGMENTED GENERATION PIPELINE FOR CONTEXT-AWARE MEDICAL QUESTION ANSWERING

The increasing demand for reliable, context-aware medical information has led to the emergence of intelligent conversational systems in healthcare. This paper introduces Med Sync-AI, a Retrieval-Augmented Generation (RAG) based medical assistant designed to provide factually grounded re- sponses to clinical and general health-related queries. The pro- posed pipeline integrates Lang Chain, Hugging Face Sentence Em- beddings, and Pinecone Vector Databases for efficient document retrieval, while leveraging Ollamas Mistral model for contextual response generation. The system dynamically reformulates user queries using a contextualization module to preserve conversation continuity, followed by a retrieval and synthesis process guided by medical literature. Quantitative evaluation of top-k similarity scores demonstrates robust retrieval precision, and qualitative analysis highlights Med Sync-AIs ability to produce concise, evidence-based answers. Overall, this framework provides a step toward developing trustworthy, transparent, and explainable AI- driven medical dialogue systems.