Medical Chatbot Development: Leveraging Artificial Intelligence to Enhance Health Education Through the Gale Encyclopedia of Medicine, OpenAI Embeddings, and Pinecone Vector Storage

Abstract

The proliferation of digital health technologies has created both an opportunity and an imperative to deliver accurate, accessible medical information to the general public. This paper presents the design, implementation, and evaluation of a Medical Artificial Intelligence Chatbot—an open-source, conversational system grounded in the Gale Encyclopedia of Medicine. The system employs OpenAI language model embeddings for semantic understanding and Pinecone vector storage for high-efficiency document retrieval. A Flask-based web interface enables end-users to pose natural-language medical questions and receive contextually appropriate responses in real time. The proposed architecture integrates PDF-based document ingestion, chunk-level text splitting, dense vector indexing, and retrieval-augmented question answering into a cohesive pipeline. Experimental evaluation across diverse medical queries demonstrates the system’s capacity to provide accurate, actionable information on topics including hypertension management, diabetes care, and respiratory symptom recognition. Performance limitations observed in highly specialized or ambiguous queries motivate a set of targeted recommendations for future improvement, encompassing domain-specific fine-tuning, ontology integration, and user-feedback-driven refinement.