The test detected no early SOZ AUC gain for CAR gamma change in 25 SEEG patients (0–5 versus 20–25 seconds; independent patients; sign symmetry)

We tested gamma power change at 0 to 5 seconds and 20 to 25 seconds after source seizure onset. The sample had 25 public SEEG patients, two seizures per patient, and 2,932 matched contacts. The score used a common average reference (CAR), 30 to 80 Hz power, and a fixed ten-second baseline. It excluded 55 to 65 Hz. Mean source SOZ label AUC was 0.628 early and 0.649 late. The mean early-minus-late difference was -0.021. Its 95 percent patient bootstrap interval was -0.112 to 0.076. The planned two-sided patient sign-flip test gave p = 0.6668. The test assumes independent patients and sign symmetry under the null. It did not detect an early gain. The interval permits a difference in either direction. This does not establish equal performance or a later advantage. All planned method checks also had intervals that included zero. The package preserves the fixed plan, code, source metadata, per-second powers, contact scores, patient results, two raw examples, and retrieval instructions. Independent signal and AUC checks passed. A second source retrieval rebuilt all features exactly. This is a new analysis of a prior AutoSci sample, with clinical source labels and no independent patient validation.

stereoelectroencephalography · epilepsy · seizure onset zone · gamma power · common average reference · area under the ROC curve · seizure timing · signal processing · patient-level inference · OpenNeuro ds004100