Matched low-frequency SOZ–nonSOZ coupling deficit recurs in a later clip of the same HUP SEEG cohort

We repeated a pre-registered test in a second, non-overlapping clip from the same HUP intracranial EEG recordings (OpenNeuro ds004100, version 1.1.3; interictal SEEG run-01, EDF records 200–289). For each SOZ-to-nonSOZ inter-shaft edge, we subtracted mean 1–40 Hz coupling on nonSOZ-to-nonSOZ edges from the same shaft pair within 5 mm in length. The mean patient-level difference was −0.0087 (95% CI −0.0149 to −0.0031; 11/32 patients above zero; Holm p = 0.012), replicating the earlier clip result of −0.0075. Across 32 patients with both clips, values correlated at r = 0.93; the clip B minus clip A difference was −0.0012 (95% CI −0.0035 to +0.0012; p = 0.33). The envelope result at 65–115 Hz was not stable between clips. This is a replication over time in the same patients, files, montage, and SOZ labels, not an independent cohort. The effect is small (about 12% of mean nonSOZ-to-nonSOZ coupling) and appears only after matching shaft pair and length; without that match the raw difference was near zero. A stable montage or reference artifact remains possible.

stereo-EEG · seizure onset zone · functional connectivity · interictal EEG · replication · OpenNeuro ds004100