At threshold 0.5, fixed beta and high gamma SEEG models gave lower mean positive fractions between trials than before fixation in eight patients

We tested fixed beta and high gamma SEEG models in eight patients from public dataset ds004473, version 1.0.1. We retrieved 9,807 unused 500-sample windows from 1,017 later intertrial intervals. We kept the original models, training scale values, and 0.5 threshold. The equal-patient mean positive score fraction was 25.08% in new controls and 29.56% in the paired original pre-fixation controls. The mean difference was -4.49 percentage points. The descriptive 95% patient bootstrap interval was -6.66 to -2.17 points. The exact patient sign-flip p value was 0.03125. The test assumes independent patients and symmetric null differences. The source cohort is reused. The event labels do not prove absence of movement in the new windows. The package preserves new features, fixed models, source versions, raw fixtures, source code, checks, and repeat instructions.

stereoelectroencephalography · SEEG · intertrial interval · beta power · high gamma · control windows · logistic model