MedMind is an advanced Retrieval Augmented Generation (RAG) chatbot designed to provide reliable and informative responses to your medical inquiries.
- Answers General Medical Questions: Utilizes a curated medical knowledge base and advanced language models to answer a wide range of health-related questions.
- Summarizes Research Articles: Retrieves and summarizes relevant research articles from PubMed, making complex scientific information more accessible.
- Offers Additional Insights: Provides context and insights from its knowledge base to enhance understanding.
- Safe Web Searches: Performs safe web searches using the Google Search API, ensuring the safety and relevance of the results.
- Hallucination Evaluation: Employs a model to minimize the risk of generating fabricated information.
- Streamlit: For building the user interface.
- LlamaIndex: For indexing and querying data.
- Vectara: As a vector database for storing and retrieving medical knowledge.
- TogetherLLM: Large language model for generating responses and summaries.
- LangChain: For building question-answering chains.
- Chroma: For indexing and querying uploaded documents.
- Hugging Face Transformers: For embeddings and hallucination evaluation.
- PubMed API: For searching and retrieving research articles.
- Google Search API: For performing safe web searches.
- Not a substitute for professional medical advice.
- Information is for general knowledge and educational purposes only.
- Under development and may occasionally provide inaccurate or incomplete information.
- Potential for hallucinations, especially for complex queries.
- Clone the repository:
git clone https://github.com/your-username/MedMind.git - Install dependencies:
pip install -r requirements.txt - Set up environment variables: Refer to the
.env.examplefile and create a.envfile with your API keys. - Run the Streamlit app:
streamlit run medmind.py
https://huggingface.co/spaces/jayash391/RAG_MedMind
Contributions are welcome!
This project is licensed under the MIT License.