/MediMatch-Clinical-Symptom-Mapper-and-Specialist-Recommender

AI-powered disease detection using patient descriptions and doctor recommendations. This initiative is tailored to the context of Bangladesh.

Primary LanguageJupyter Notebook

How to run Backend model

  1. Run backend/models/identify-symptoms-using-huggingface-transformer.ipynb, it'll create a fine-tuned-model folder that will contain the model details

  2. Then simply run the backend uvicorn main:app --reload and navigate to Sweeger UI

  3. Test the get_symptoms api to extract symptoms from natural language.

Resources

Doctors Image: https://drive.google.com/drive/folders/19drVDiuWjgQXLOfSosXugfibPXGQ0yVH?usp=drive_link

TODO:

Currently we are using Huggingface Transformer to extract symptoms from text which is supervised learning method.

Next if possible we can try to do some semi-supervised learning for achieveing better accuracy