SEP 9, 2026 · PREPRINT
NEXUS-MI: Communication-Aware Federated Personalization for Gateway-Coordinated Motor-Imagery Brain-Computer Interfaces
arXiv
This is a computational methods paper demonstrating a federated learning framework for BCI personalization using offline replay on two public datasets, without clinical validation, real-time deployment, or comparison to standard clinical BCI practice.
Reported
Server-to-client backbone traffic…approximately 42%
Maximum subject-level accuracy lo…approximately 12 percentage points
Sample size (BCICIV-2a)9