80 to 150 Hz power order gave no clear onset label AUC gain in 16 ds004100 SEEG test patients. The test used fixed linear models and assumed independent symmetric null signs.

We tested whether the order of high gamma power adds clinical onset label information beyond typical power, concentration, and contact number. The source was public SEEG from OpenNeuro ds004100 version 1.1.3. We used two 60-second records per patient and a fixed split of 17 training patients and 16 test patients. Mean within-shaft test AUC rose from 0.78463 to 0.79549. The change was 0.01086, with a patient bootstrap 95 percent interval of [0.00009, 0.02551]. The exact two-sided sign-flip test gave p = 0.09570. It did not reject zero at 0.05 under independent patient differences with symmetric null signs. The interval holds training fits fixed. One patient supplied about 58 percent of the summed AUC change. Power order rank agreement between records was weak: mean 0.08043. All six fixed secondary checks are reported. The package preserves code, 70 input power files, metadata, raw source version locks, predictions, results, and repeat commands. All 70 inputs were rebuilt from fresh pinned raw prefixes. This same-cohort test does not establish an AUC gain, surgery benefit, or performance at a new center.

SEEG · intracranial EEG · high gamma power · feature order · seizure onset zone · AUC · epilepsy