SwarmMed: an open-source platform for decentralized, collaborative and privacy-preserving biomedical data analysis
Year: 2026
Type: software
Abstract: SwarmMed is an open-source platform for running decentralized (swarm) learning studies on biomedical data across institutions. It provides a web interface (SwarmMedHub) and a command-line interface (SwarmMedCLI) over the NVIDIA FLARE client-controlled swarm workflow, covering project and consortium setup, network provisioning and startup-kit distribution, site-local data registration and sandboxed validation, decentralized training, and results synthesis. Sites exchange model parameters and training metadata; no message sent by SwarmMed or NVIDIA FLARE contains raw data.
Institutions: Heidelberg University, University Hospital Heidelberg, National Center for Tumor Diseases, University Hospital Carl Gustav Carus, Else Kröner Fresenius Center for Digital Health, Technische Universität Dresden, Fraunhofer Institute for Digital Medicine, German Center for Neurodegenerative Diseases, Purdue University West Lafayette
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225315184