In two SEEG patients, sleep coupling did not improve held-out CCEP prediction, despite a small low-frequency relation in one patient

We tested whether spontaneous coupling in a separate 240 s sleep recording predicts cortico-cortical evoked potential (CCEP) strength beyond distance in the two SEEG patients from OpenNeuro ds003848 v1.0.3 who had both SPES and sleep runs. The analysis included 71 stimulated sites and 4,458 site-channel pairs. Low-frequency (1–40 Hz) coupling had a small positive mean partial Spearman relation to log CCEP strength after control for log distance (rho = 0.061, sign-flip p = 0.005; 46/71 sites positive), driven by RESP0749 (rho = 0.123, p = 0.002); the estimate was small and not significant in RESP0800 (rho = 0.020, p = 0.40). High-frequency (70–145 Hz) envelope coupling showed no relation (mean rho = 0.002, p = 0.90). In leave-one-site-out prediction, adding coupling to distance reduced mean held-out Spearman rho from 0.180 to 0.157 (paired p = 0.075) and responder AUC from 0.789 to 0.774 (paired p = 0.23). Thus, coupling did not improve prediction in these data. The study includes only two patients, and sites within a patient are not independent, making the pooled p-value optimistic. Sleep stage was not scored, the sleep data were not visually reviewed, and stimulation artefact was not removed; the distance exclusion and analysis window reduce but do not eliminate that risk.

cortico-cortical evoked potentials · single-pulse electrical stimulation · sleep · SEEG · functional connectivity · distance · epilepsy