Under patient independence, the onset AUC change interval for added baseline gamma included zero in a fixed common average SEEG model with 16 test patients

We tested if baseline gamma power adds useful information to early seizure gamma power for contact ranking. The source was public OpenNeuro ds004100, version 1.1.3. We used two seizures from each of 32 patients with stereoelectroencephalography (SEEG). We fit fixed models on 16 patients and tested them on the other 16, with no patient in both sets. Mean test-patient onset AUC was 0.7981 for early power alone and 0.7862 for early plus baseline power. The mean paired change was -0.01184 AUC units; its 95 percent patient bootstrap interval was -0.02889 to +0.00370. The exact two-sided patient sign-flip test gave p = 0.19653 and did not detect a gain. The interval holds fitted models fixed and assumes independent test patients; the sign-flip test also assumes null sign symmetry. This does not establish that baseline power has no information. Six test patients had a positive change, eight a negative change, and two no change. The local plan preceded the new scores, but prior univariate results from this cohort were known, so this was not a blind or external-cohort test. A planned supplied-signal check had a positive change. The result is limited to the fixed primary model, settings, and selected cohort.

stereoelectroencephalography · SEEG · gamma power · seizure onset · contact ranking · patient-level test · baseline feature · AUC