Passive coupling adds little to distance for N1 CCEP prediction in eight SEEG patients

We tested whether passive coupling between stereoelectroencephalography (SEEG) contacts predicts expert-checked N1 cortico-cortical evoked potentials (CCEPs) beyond distance and same-shaft status. In eight patients from OpenNeuro ds005448 v1.0.0, we measured high-gamma (70–150 Hz) envelope correlation and low-frequency (4–30 Hz) signal correlation in 30 two-second windows per patient, taken 1.5–3.5 s after 1–2 mA pulses in the first SPESclin run. For response contacts at least 10 mm from both stimulated contacts, patient-specific ridge logistic models used five-fold cross-validation grouped by stimulation site, repeated 10 times. The baseline used log distance and same-shaft status; the second model added both coupling features. Mean AUC gain was 0.014 (median 0.007), positive in five of eight patients; the exact one-sided sign-flip p-value was 0.055. The pre-set rule (p < 0.05 and at least six patients with positive gain) was not met. In a secondary analysis without correction for two tests, adding high-gamma coupling alone gave a mean gain of 0.015 (six of eight patients, p = 0.016); low-frequency coupling alone gave 0.001 (two of eight, p = 0.375). Distance remained the main predictor (baseline cross-validated AUC 0.70–0.92). These results do not show that passive coupling can replace test stimulation. Windows followed stimulation and were not true rest; the analysis included 60 seconds per patient and eight patients from one center.

stereoelectroencephalography · SEEG · cortico-cortical evoked potentials · CCEP · N1 · passive coupling · functional connectivity · high-gamma · OpenNeuro · prediction