In eight SEEG patients, spontaneous coupling did not improve N1 CCEP prediction beyond distance (mean AUC gain 0.0001)

We tested whether zero-lag spontaneous coupling predicts expert-checked N1 responses to electrical stimulation (CCEPs) beyond distance in eight SEEG patients from OpenNeuro ds005448 v1.0.0. Across 26,052 pair-contact rows from 281 stimulation pairs, coupling was measured in about 80 seconds of baseline and inter-pulse data per patient. Coupling alone ranked responders above chance (mean AUC 0.69 for 1–40 Hz signals and 0.64 for 70–150 Hz envelopes; 8/8 patients above 0.5), but it decreased with distance (Spearman rho −0.18 to −0.39). Adding the primary 1–40 Hz feature to log distance changed mean grouped five-fold cross-validated AUC from 0.8535 to 0.8536 (gain +0.0001; 95% CI −0.0013 to 0.0014; exact sign-flip p = 0.82; 4/8 patients positive). The largest mean gain across secondary group analyses was 0.0046 for the 70–150 Hz feature among contacts at least 15 mm away (p = 0.11); its leave-one-patient-out gain was −0.0009 (95% CI −0.0015 to −0.0004; p = 0.008). These results are limited to one run per patient, short recordings dominated by inter-pulse windows, binary expert N1 labels, and zero-lag Pearson coupling. Late stimulation effects may remain in the inter-pulse windows; longer resting data and lagged measures were not tested.

SEEG · CCEP · N1 response · brain connectivity · electrical stimulation · distance · AUC · epilepsy