In 12 SEEG Patients, Nearby Pre-train 70–150 Hz Power Did Not Predict Cortical-Stimulation Effect Current

Can spontaneous SEEG near a cortical stimulation site identify sites that need less current? We examined 5 s of pre-train activity from OpenNeuro ds005169 v1.0.0 (visual percepts; 112 recordings) and ds006519 v1.0.0 (negative motor responses; 35 recordings), using HTTP byte ranges. The public recordings lacked the stimulated bipolar pair, so we measured same-shaft adjacent bipolar pairs within one contact index of a stimulated contact. These flank pairs were available in 106 of 147 recordings. The prespecified primary test was the within-patient Spearman relation between relative 70–150 Hz log power and log current in patients with at least three recordings and at least two distinct currents. In 12 patients and 81 ds005169 recordings, mean rho was −0.11 (95% patient-bootstrap CI −0.32 to 0.11; exact two-sided sign-flip p = 0.43); 7 of 12 patient correlations were negative. No relation was found. Exploratory relative 13–30 Hz power had mean rho −0.36 (95% CI −0.54 to −0.18; p = 0.006 uncorrected; Holm-adjusted p = 0.035 across six relative bands; Bonferroni p = 0.07 across 12 band-by-normalisation tests). The pooled patient-centred analysis across 97 recordings gave rho = −0.25 (p = 0.013) for 13–30 Hz. This beta-band result is a hypothesis for confirmation. The feature was measured near, not at, the stimulated pair; effect current is not a measured threshold. Few recordings per patient, tied currents, no tissue or anatomy covariate, possible patient overlap across datasets, single-centre data, and offline analysis limit interpretation. These findings do not support setting a write-BCI start current from nearby pre-train high-gamma power.

SEEG · intracranial EEG · cortical stimulation · high-gamma power · beta power · effect current · within-patient analysis · neural engineering