BCIwiki (bciwiki.com) — The study was published in Computers in biology and medicine on June 26, 2026. The researchers introduce LEGEND, a neural-decoding architecture that jointly processes cortical EEG, spinal ESG, and peripheral-muscle EMG. The aim is to model the motor hierarchy from cortical intent to muscle activity after spinal cord injury within a single framework.
LEGEND encodes the three modalities in Lorentz hyperbolic space, connects 51 channel nodes through a signed tri-layer phase-locking-value graph, and refines the representation with graph attention. Under strict leave-one-subject-out evaluation on the Steele dataset, the model achieved 56.51%±12.27% accuracy, 23.4 percentage points above EEGNet. The researchers argue that hyperbolic representations help capture complex relationships across the motor hierarchy and can provide a computational basis for neural bypasses and rehabilitation decoding that link brain, spinal, and muscle activity.