RAG-Based Patient Review Question-Answering Chatbot

An AI system that helps hospitals instantly analyze patient reviews and answer operational queries using retrieved real-world feedback.

Github

LLM & Generative AI

• Google Gemini 2.5 Flash (Chat Model)
• LangChain (Prompt templates, Runnables, Chains)
• Google Generative AI Embeddings (gemini-embedding-001)

Vector Database

• ChromaDB (persistent vector store)
• Similarity search retriever (k=10)

Data Handling & Preprocessing

• CSVLoader (for ingesting reviews.csv)
• Batched embedding generation with rate-limit safety
• Document structuring for retrieval modules

Prompting & RAG Pipeline

• ChatPromptTemplate
• System + Human message templates
• Context injection for hallucination-safe answers
• StrOutputParser for clean output formatting

Frontend / UI

• Gradio ChatInterface
• Real-time question answering
• Chat history & session-based interaction

Others

• Python, Colab, time-based batching, reproducible pipelines