Developing a RAG-Based Chatbot for Healthcare: A Case Study of the HeartWise AI Chatbot

HeartWise is a proof-of-concept Retrieval Augmented Generation (RAG) chatbot supporting self-care in coronary artery disease. Eight simulated scenarios were used to assess preliminary performance. Six cases (75%) produced guideline-consistent responses, while two vague medication-related queries exposed limitations in context handling. Iterative refinements to the knowledge base, retrieval settings, and prompt engineering improved consistency. Given the small, qualitative evaluation, findings indicate early feasibility, not intended for clinical decision making. The study outlines practical design considerations for transparent, safe RAG applications in healthcare.