MediBot Chat

MediBot Chat is an intelligent medical question-answering assistant built with retrieval-augmented generation (RAG) and local LLaMA 3 language model integration via Ollama. The app uses FAISS and SentenceTransformers to search a custom medical dataset and Streamlit for an interactive chat UI.

---

Features

• Semantic search over medical transcripts using FAISS
• Local LLaMA 3 integration with Ollama for natural language generation
• Streamlit-based chat interface with conversation memory
• Custom prompt engineering tailored for medical advice
• Easy setup and extensible design

---

Installation & Setup

1. Clone the repository

git clone https://github.com/aireynexxx/MediBot.git

2. Install Requirments

pip install -r requirements.txt

3. Prepare embeddings

Make sure you have your medical dataset CSV file in data/medidata.csv.

Run the embedding script to create FAISS index and metadata:

python embedder.py

4. Start the Streamlit app

streamlit run app.py

Usage

• Type your health-related question in the chat input.

• MediBot will retrieve relevant context and generate a medically plausible answer.

• Use the sidebar button to clear chat history if needed.

Credits

• UI inspired by the Streamlit official tutorial: How to Build an LLM-Powered Chatbot
• Data used: Medical Transcriptions