Pairs selected for low beta change in one cue condition had lower changes than other pairs in the other cue condition in eight SEEG patients

Pairs chosen for low beta power change in one cue condition had lower test changes in the other condition. We tested 512 sampled keypress trials and 819 adjacent electrode pairs in all eight patients from OpenNeuro ds004473, version 1.0.2. For each test group, separate trials from the other cue condition selected the lowest 20 percent of pairs. Training and test intervals had a gap of at least 30 seconds. The contrast was the selected-pair mean minus the remaining-pair mean. Both conditions and all patients had equal weights. All eight patient contrasts were negative. The mean contrast was -0.919 dB. The 95 percent patient bootstrap interval was -1.512 to -0.434 dB. The exact two-sided patient sign flip test gave p = 0.0078125. This test assumes independent patients and symmetric effects under the null hypothesis. Selected pairs had a mean test change of -0.509 dB. The lowest-20-percent lists for the two whole-condition maps shared 36.8 percent of their pairs on average. Thus, transfer of a relative signal did not imply equal site lists. Checks of selection size, time gap, reference, and a later amplitude filter gave the same contrast direction. Only three patients met the prior-training-only rule. The primary test was offline and could use later blocks. Condition and time can be associated. Raw beta power can include aperiodic activity and artifacts. This same-cohort test does not establish motor specificity, clinical accuracy, or transfer to new sessions or patients. The package preserves source code, input powers, exact trial splits, selected sites, all uncertainty draws, and a public retrieval recipe. All used signal ranges were downloaded again and gave the same powers. Separate calculations and an offline rerun passed.

stereoelectroencephalography · beta power · condition transfer · site selection · intracranial EEG · movement