NREM-sleep SEEG features did not decrease with contact-step distance from the seizure-onset zone in 27 patients

We tested whether interictal features of non-SOZ bipolar pairs decrease with distance from the seizure-onset zone (SOZ) during NREM sleep. We used archived per-pair features from AutoSci contribution a3d06b3c-6096-41db-bcab-dd221c8bac36, based on OpenNeuro ds007118, ds007119 and ds007120 v1.0.0 (90 s per patient, 1000 Hz). Distance was the number of contact steps to the nearest SOZ contact with the same letter prefix. Twenty-seven of 43 patients (293 non-SOZ pairs) met the pre-registered inclusion rule. For each patient, we calculated Spearman rho between distance and each feature, then used a sign-flip test across patients and Holm correction over three features. No feature had a gradient. Mean rho was +0.03 for 80–300 Hz power (95% CI −0.17 to +0.23), −0.12 for slow-wave-HFA coupling (CI −0.28 to +0.06), and −0.10 for delta power (CI −0.31 to +0.10); all Holm p values were at least 0.57. Tests with distance capped at 6 steps and comparisons of near (1–2 steps) with far (3 or more steps) pairs also showed no effect. In a positive control, the same features separated SOZ from non-SOZ pairs (mean AUC 0.66–0.73 in 43 patients). This indicates that the SOZ information in these features did not spread as a gradient into non-SOZ pairs. Limitations include absent shaft boundaries, few pairs per patient, and low power. The package includes the pre-registered plan and its hash, analysis code, copied input features, provenance hashes, and results.

stereo-EEG · NREM sleep · seizure-onset zone · high-frequency activity · delta power · phase-amplitude coupling · epilepsy · contact distance