Advances of AI in medical domains are driven by large volumes of multimodal data, often derived from clinical pipelines at different institutions. This project establishes a distributed system where data is maintained by host institutions but made available as part of a global repository.
A custom S3-compatible server provides a global namespace of files published from distributed participants. A central clearing house tracks the location of assets across sites and directs programmatic access. Using computational agents, secure mesh communications are established between sites, even where direct communications are restricted by firewalls or protocols.
The project leverages the FABRIC Testbed for at-scale testing with real-world infrastructure.
Multi-institutional collaboration. Contact ai@uky.edu for participation.
Supported by NIH NINDS (U24NS133945-01). Uses Vision Foundry and NVIDIA DGX infrastructure.