Skip to main content

Federated Learning: Diabetes Prediction

Work in Progress

This tutorial is being updated for syft-flwr. Check back soon!

Overview​

Learn how to build a privacy-preserving diabetes prediction model using federated learning with syft-flwr. This tutorial demonstrates how multiple data owners can collaboratively train a machine learning model without sharing their sensitive health data.

What You'll Learn​

  • Setting up FL clients (member nodes) with local diabetes datasets
  • Configuring an FL server (aggregator) for model coordination
  • Training a logistic regression model across distributed datasites
  • Implementing privacy-preserving aggregation strategies
  • Evaluating federated model performance

Tutorial Structure​

This tutorial is divided into two main parts:

  1. Part I: Data Owner - Configure data owner nodes to participate in federated training
  2. Part II: Data Scientist - Set up the aggregator to coordinate federated learning rounds

Prerequisites​

  • SyftBox client installed and configured
  • Python 3.11+
  • Basic understanding of machine learning concepts
  • Local or synthetic diabetes dataset

Next Steps​

Continue to Part I to set up your FL client and begin participating in federated model training.

Complete tutorial content coming soon!