Helping healthcare professionals make faster, research-backed clinical decisions with Gen AI + RAG
MediInsight Health Solutions is a US-based healthcare provider that helps doctors and research assistants deliver clear, evidence-based advice to patients.
When a patient asks about a medical intervention — such as intermittent fasting for Type 2 diabetes — the scientific evidence can be scattered, inconclusive, or hard to retrieve quickly.
To solve this, we developed MediAssist AI — a Retrieval-Augmented Generation (RAG) system that automatically:
- Fetches medical research from PubMed using APIs.
- Embeds & stores articles in a vector database for fast semantic search.
- Retrieves relevant evidence for a query.
- Generates clear, contextual answers using the Groq LLaMA 3 model.
This empowers healthcare professionals to save time, reduce uncertainty, and base recommendations on the latest peer-reviewed research.
- Automated PubMed Data Pipeline:
- Search and fetch the latest research articles for any medical topic.
- Store articles in JSON format for reproducibility.
- Vector Database with ChromaDB:
- Store embeddings for semantic similarity search.
- Optimized for biomedical text with the S-PubMedBert-MS-MARCO model from Hugging Face.
- RAG Pipeline with Groq LLaMA 3:
- Retrieve the most relevant research context.
- Generate evidence-based responses.
- Streamlit Web App:
- Sidebar to search & ingest articles.
- Main area for natural language medical queries.
- Option to manage (clear/delete) collections.
- Modular Design:
- Easily swap embedding models or LLMs.
- Environment variables stored securely in
.env.
https://mediassist-genai.streamlit.app/
-
Clone the Repository
git clone https://github.com/MindMatrixPro/MediAssistAI.git cd MediAssistAI -
Install Dependencies
pip install -r requirements.txt
-
Run the Streamlit App
streamlit run main_app.py
To Contribute, please submit issues or pull requests for enhancements or fixes.
Licensed under the Apache 2.0 License.