NREM slow waves co-occur across SEEG shafts above chance, but absolute involvement varies with detection settings

We tested whether non-rapid eye movement (NREM) slow waves in human intracranial EEG are local or global, using shaft identity as a proxy for cortical region. We analyzed 43 archived 90-s windows from OpenNeuro ds007118, ds007119, and ds007120 (snapshot 1.0.0); the 15 patients meeting the multi-shaft inclusion rule were all from ds007119. The sample included 288 adjacent-contact bipolar pairs and 15,271 detected slow waves. With the planned 0.5–4 Hz, top-20% detector and ±0.4-s co-occurrence window, involvement exceeded a circular-shift chance estimate by 0.142 (95% CI 0.093–0.197; all 15 patients; Holm-adjusted p=0.0003). Seed amplitude was positively associated with between-shaft involvement (mean Spearman rho 0.173, 95% CI 0.116–0.233; 14/15 patients; Holm-adjusted p=0.0003). Within-shaft involvement exceeded between-shaft involvement by 0.060 (95% CI 0.035–0.087; 12/13 patients; uncorrected p=0.0004). The planned test of median involvement below 0.5 was not significant with the main settings (mean difference +0.010, 95% CI −0.093 to +0.115; p=0.86). It was significant in all three robustness variants, while the excess above chance remained positive (0.133–0.156). Thus the absolute involvement fraction was not stable across detection windows and thresholds, but co-occurrence above chance and its association with amplitude were consistent. The analysis used only 90 s per patient, from one dataset for the inferential sample; patients had epilepsy, shaft locations lack coordinates, and bipolar derivation can reduce far-field signals.

NREM sleep · slow waves · stereo-EEG · SEEG · bipolar EEG · cortical involvement · co-occurrence · epilepsy