Pre-press SEEG power decoded cued key side in one of two patients (AUC 0.83); cue and key sides were confounded

This analysis asked whether bipolar stereoelectroencephalography (SEEG) band power in the 0.5 s before a cued key press classified left versus right key presses in two patients from OpenNeuro ds004473 v1.0.2. Using response-locked beta (13–30 Hz) and high-gamma (70–170 Hz, excluding 115–125 Hz) power, shrinkage LDA, stratified five-fold cross-validation, and label permutations, combined beta and high-gamma power reached AUC 0.83 (permutation p = 0.005, the minimum) in sub-2, but AUC 0.44 (p = 0.82) in sub-1. At the earliest tested time window, sub-2 AUC was 0.77, approximately 0.3 s after the go cue. A response-window positive control had AUC 0.97–0.98 in both subjects. The result is subject-dependent and does not support a population claim. Cue side and pressed-key side agreed in 98.0% of sub-1 and 100% of sub-2 trials, so this analysis cannot distinguish a cue-driven signal from motor planning. The hand used for each key was unknown. Only two of the dataset's subjects were analyzed; shrinkage was fixed, and the time-resolved and control analyses were added after the first sub-1 result.

stereoelectroencephalography · SEEG · brain-computer interface · motor planning · decoding · beta power · high-gamma power · OpenNeuro