Clusters of the six largest high frequency power windows showed no clear label difference in 19 ds003876 SEEG patients. Tests assumed independent symmetric patient contrasts.

We repeated a test of large power window clusters in an external SEEG cohort. We used 19 patients from OpenNeuro ds003876, version 1.0.2. We selected the six largest one-second powers in the 80–150 Hz band, with 118–122 Hz excluded. The cluster score counted selected neighbors. The local comparison used 77 marked and 254 unmarked pairs in 45 contact name groups. Unmarked contacts had source value n/a. Groups, clips, and patients had equal weight at each level. The mean marked minus unmarked difference was +0.59 percentage points. The conditional 95 percent patient interval was −3.38 to +4.97 points. The exact two-sided test gave p = 0.795 under independent symmetric patient contrasts. The local AUC was 0.486. A descriptive three-window check gave a negative difference of −2.80 points, with interval −5.27 to −0.16 points. Fresh pinned source ranges reproduced all 408,240 power values exactly. Separate calculations and a full offline repeat matched the results. The package preserves code, inputs, metadata, source recipes, results, and instructions. The primary result does not establish equal scores or clinical benefit.

stereoelectroencephalography · seizure onset zone · high frequency power · time clustering · patient-level inference · OpenNeuro