Shared SEEG power correction gave a 0.009 local AUC timing contrast in 30 patients with two 60 s interictal clips. Its 95 percent interval included zero.

We tested whether the time alignment of a shared power correction changes local seizure-onset label AUC. We used public OpenNeuro ds004100 version 1.1.3 and two 60 s interictal SEEG clips per patient. The correction used centered log powers from other electrode name groups. Clinical labels did not select its inputs. The score was the fraction of 80–150 Hz power in the six largest one-second windows, with 118–122 Hz omitted. We compared each positive pair with negative pairs in its own name group. Thirty patients supplied 196 positive and 341 local negative pairs. Aligned correction gave a mean local AUC of 0.7230. The mean over 59 nonzero factor shifts gave 0.7142. Uncorrected AUC was 0.7203. The paired timing difference was 0.0089, with a 95 percent patient bootstrap interval of [-0.0034, 0.0217]. The planned patient sign flip test gave p = 0.1809. It assumes independent patients and symmetric patient differences under a zero null. The five fixed checks also gave timing intervals that included zero. This test gives no clear evidence of a mean timing effect under its assumptions. It does not prove equivalence or identify the source of shared activity. All 72 pinned source prefixes matched their saved hashes. All 408 rebuilt target power arrays matched the earlier package exactly. Independent calculations and a complete offline repeat passed. The package preserves code, source identities, label-independent powers, correction factors, shifts, scores, patient results, random draws, figures, and reproduction commands.

stereoelectroencephalography · SEEG · epilepsy · seizure-onset zone · high-frequency power · power concentration · signal correction · time-shift control · area under the curve · reproducibility