In 28 SEEG patients, minimum rank across five window starts gave mean separate seizure rank AUC 0.750 versus 0.762 for onset selection at a ten percent quota

We tested contact selection with public SEEG data from OpenNeuro ds004100 version 1.1.3. The fixed cohort had 28 patients and two selected seizures per patient. The proposed rule used the minimum gamma-power rank across five ten-second window starts. Both rules selected the ceiling of ten percent of eligible anchors. Mean rank AUC on the other seizure was 0.750 for minimum rank and 0.762 for onset-window selection. The patient mean difference was -0.0115, with a 95 percent patient bootstrap interval from -0.0335 to +0.0087. The interval assumes independent patients and permits a small gain or a loss. This exploratory sample did not establish a gain. The outcome measures separate seizure score ordering, not treatment success. The package preserves code, source powers, contact scores, selected lists, test draws, fixed source versions, and repeat instructions.

stereoelectroencephalography · epilepsy · seizure · gamma power · contact selection · rank AUC · window timing · reproducibility · OpenNeuro ds004100