A phase decoder for reconstructing missing muscle activity

A two-stage phase-decoder reconstructs missing muscle activations from sparse EMG in neurologically intact participants (N=4), achieving median movement-phase RMSE <0.17 and activation RMSE 0.08–0.34.

• Why it matters: Sparse usable EMG from nerve injury, muscle loss, or limited electrode placement constrains decoding of intended upper-limb movements for assistive control.
• What they did: The method first estimates ongoing movement phase from available EMG, then retrieves matching muscle activations from a library of pre-recorded templates in real time. Proof-of-concept tests used object-transfer tasks under simulated denervation.
• The result: Movement phase estimation had median RMSE <0.17 of movement duration, and reconstructed missing activations reached RMSE 0.08–0.34 with correlation up to 0.97 within ~2 ms, including in a previously unseen participant.

AI summary of the abstract · may contain errors · not medical advice · read the paper.