Under independent patient sampling, the 95 percent bootstrap interval included zero for shared-contact retention gain of top 10 percent 1–80 Hz SEEG hubs (36 patients).

We tested hub agreement between two 60-second 1–80 Hz SEEG samples from each of 36 patients with epilepsy. We used public OpenNeuro ds004100 data. Hubs had the largest mean absolute correlations. The full graph retained 68.6 percent of its top 10 percent hubs. The graph without shared-contact edges retained 66.7 percent. The paired mean difference was 1.90 percentage points. Its 95 percent patient bootstrap interval was −3.07 to 8.95 points. The interval assumes independent patient sampling. It does not establish zero effect or graph equivalence. A model with zero covariance between different contacts retained 81.7 percent of its hubs. It kept the recorded variances and montage. Model hubs had only 6.14 percent overlap with observed hubs. The package preserves covariance inputs, code, patient results, model checks, and public retrieval recipes. The short clinical samples and variable sensitivity results limit general claims.

SEEG · hub retention · bipolar channels · shared contacts · connectivity · patient bootstrap