RAGMed

RAGMed leverages Retrieval-Augmented Generation (RAG) to analyze medical question-and-answer datasets, providing users with expert-informed recommendations for their medical queries. This project aims to enhance medical decision-making by integrating advanced natural language processing techniques with reliable medical data sources.

Features

• Retrieval-Augmented Generation (RAG): Combines retrieval-based methods with generative models to produce accurate and contextually relevant responses.
• Medical Q&A Analysis: Processes medical question-and-answer datasets to extract pertinent information and deliver expert-informed recommendations.
• User-Friendly Interface: Designed to be accessible for both healthcare professionals and patients seeking medical information.

Installation

1. Clone the repository:

   git clone https://github.com/BijanProjects/AI_Medical_Agent.git
   cd AI_Medical_Agent

2. Create and activate a virtual environment:

   python -m venv env
   source env/bin/activate  # On Windows: env\Scripts\activate

3. Install the required dependencies:

   pip install -r requirements.txt

Usage

1. Prepare the medical Q&A dataset:

   A simple dataset is provided: Comprehensive Medical Q&A Dataset. You can modify the Medical_QA.csv file to include your dataset. Ensure your data is in CSV format and follows the same structure as Medical_QA.csv.

2. Vectorize the dataset:

   Use the Vectorization.py script to convert textual data into vector representations suitable for retrieval tasks.

3. Initialize the retriever:

   Utilize the Retriever.py script to set up the retrieval system that will fetch relevant information based on user queries.

4. Generate responses:

   Employ the prompt_gen.py script to create prompts for the generative model, facilitating the production of expert-informed recommendations. Feel free to change the models using Ollama or Unsloth libraries (recommended).

Contributing

Contributions are welcome! Please fork the repository and create a pull request with your proposed changes.

Acknowledgments

Dataset link: Comprehensive Medical Q&A Dataset

Note:

Model is under final testing and will be shared on HuggingFace soon.