Development of a Medical Chatbot Using Generative AI and Retrieval-Augmented Techniques

Quick and accurate access to medical information is crucial for patients and healthcare professionals alike. Traditional search engines often fail to concise and context-specific responses to medical queries. This paper introduces a medical chatbot that combines large language models (LLMs) with retrieval-augmented generation (RAG) techniques to generate reliable and contextually relevant answers. The system integrates the Mistral-7B-Instructv0.3 model hosted on Hugging Face and uses Pinecone, a vector database, for efficient semantic search. Built using the Flask web framework, the chatbot retrieves the most relevant medical texts, constructs dynamic prompts, and generates accurate, user-friendly responses. Experiments with common medical questions demonstrate the chatbot’s ability to deliver timely, medically sound answers, highlighting the advantages of combining retrieval mechanisms with generative models in healthcare applications.