No detectable time-on-task change in the SEEG movement response or fixed-decoder AUC in eight subjects

We examined whether the movement-related SEEG response or rest baseline changed during one offline cued key-press session in eight subjects from OpenNeuro ds004473 v1.0.2. For each subject, eight bipolar pairs were selected using trials 1–100, and measurements were made on trials 101 to the end (5–29 min). An exact sign-flip test found no detectable time-related change in the 70–150 Hz move-minus-rest response (median slope −0.0053 log power per 10 min, −2.0% of the mean response; 4/8 negative; p = 0.34), the 70–150 Hz rest baseline (p = 0.55), the 13–30 Hz decrease (p = 0.37), or its rest baseline (p = 0.69; Holm-adjusted p = 1.0). A shrinkage LDA trained on trials 1–100 showed a median late-minus-early AUC difference of +0.013 (p = 0.78). With only eight subjects, these tests have low power; individual 70–150 Hz response slopes ranged from −0.31 to +0.05 log power per 10 min, and percent slopes are unstable when mean responses are small. The data therefore do not show a need for gain or offset adaptation within one session, but do not exclude changes in individual subjects.

SEEG · time on task · movement response · key press · band power · baseline drift · shrinkage LDA · sign-flip test · OpenNeuro ds004473