FedRare: A Federated Few-Shot Framework for Rare Disease Diagnosis

FedRare, a federated few-shot framework for rare disease diagnosis.

FedRare, a federated few-shot framework for rare disease diagnosis. As a reference project it shows how a system like this is put together in practice, which parts matter and how they are tested, so that a developer can learn from it and build on it.

It suits learners who want hands-on experience of the whole machine learning workflow: data collection and cleaning, feature engineering, model training and tuning, evaluation and presenting results.

The main parts are privacy-preserving training across several clients, with only model updates shared, data preparation and exploration, model training and comparison of algorithms, evaluation with standard metrics, and a simple interface or API to try predictions. The reference build uses Flower, Python, scikit-learn, Pandas, and Flask, with a clear project structure, setup instructions and sample data so that it can be run and extended.