A RAG-Based Health Chatbot for Verifiable Medical Queries

The research demonstrates that the Retrieval-Augmented Generation (RAG) method enables developers to create AI health chatbots that provide accurate information through transparent operations while keeping expenses at manageable levels. The system achieves its ability to create medical-based responses through the integration of the all-MiniLM-L6-v2 sentence transformer with ChromaDB vector database and Groq LLM API which operates without needing expensive hardware or complicated systems. The system uses stateless design elements to safeguard user information while its emergency detection system functions as an essential security measure. The system benefits from enhanced transparency through its method of providing source citations with every response which lets users check the supplied information. The system operates in educational institutions and rural health information centers and basic consumer health applications because it requires no special hardware and maintains low running expenses. The system demonstrates how reliable AI-based health information can become more accessible through its current implementation.

[1] P. Lewis et al., "Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks," Advances in Neural Information Processing Systems (NeurIPS), 2020. [2] N. Reimers and I. Gurevych, "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks," Proceedings of EMNLP, 2019. [3] ChromaDB Documentation, "ChromaDB Open-Source Embedding Database." [Online]. Available: https://docs.trychroma.com/ [4] Groq API Documentation, "Ultra-Low Latency LLM Inference." [Online]. Available: https://console.groq.com/docs/ [5] Streamlit Documentation, "Streamlit — The Fastest Way to Build Data Apps." [Online]. Available: https://docs.streamlit.io/ [6] K. Singhal et al., "Large Language Models Encode Clinical Knowledge," Nature, vol. 620, pp. 172-180, 2023. [7] Y. Kang et al., "A Systematic Review of Chatbots in Healthcare," NPJ Digital Medicine, vol. 6, 2023. [8] J. Devlin et al., "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding," NAACL-HLT, 2019.