MediNet is a software that allows researchers to train their own models in federated environments without programming any line of code. All the configuration is made with an easy-to-use UI.

• ✅ Train your models in federated environments
• ✅ Design models without code with our model designer (currently just sequential models)
• ✅ Your models will always train with Differential Privacy to ensure security and more anonymization
• ✅ Monitor your models during training and see how each client is performing alone, not just the global model

• 🔄 Template models: modify and create models from templates like U-Net, VGG16, etc.
• 🔄 Non-sequential model builder: We are working on creating a new model builder with the capacity to develop non-sequential models
• 🔄 More parameters to configure in your models like learning rate scheduler and other callbacks

We are currently working on improving the security measures.

• For Hospitals: If you are a hospital and want to allow researchers to access your data, please check MediNet-Node
• Documentation: Medinet Documentation
• Application Note Paper: Coming soon

• Ramon Mateo Navarro: ramon.mateo@citm.upc.edu
• Juan R. González, PhD (Supervisor): juanr.gonzalez@isglobal.org

Developed at ISGlobal