Protocol for state-based decoding of hand movement parameters using neural signals

We present a protocol for decoding kinematic and kinetic parameters from the primary somatosensory cortex during active and passive hand movements in a center-out reaching task using state-based and conventional decoders. We describe steps for preparing data and using the state-based model to classify movement directions into states via feature extraction and predict parameters with regression models (partial least squares and multilinear regression) trained per state. This state-based approach outperforms conventional methods, enhancing accuracy for brain-computer interface applications. For complete details on the use and execution of this protocol, please refer to Mirfathollahi et al.¹.