Stored low frequency SEEG scores exceeded a fixed shaft count null by 0.058 mean AUC in two 120-second recordings per patient across 36 patients

We tested a fixed low frequency SEEG score after each shaft seizure onset count was fixed. The data were OpenNeuro ds004100 version 1.1.3, through a verified AutoSci feature package. The study used 4,252 contacts and two 120-second recordings in each of 36 patients. The mean measured AUC was 0.6473. The exact shaft count null mean was 0.5892. The mean excess was 0.0581, with a 95 percent patient bootstrap interval of 0.0154 to 0.1035. No count shuffle was as extreme among 100,000 draws. The plus-one p value was 0.000010, at the simulation resolution. A cyclic label rotation check gave p = 0.000240. These tests assume the stated within-shaft label rules. Ten patient excess values were negative. The second recording patient interval included zero. The exact benchmark uses known clinical labels. The result covers short excerpts from one surgical epilepsy cohort. The package preserves code, source features, fixed source versions and hashes, all control draws, and retrieval instructions. Exact enumeration, a full second run, and one independent raw signal check passed.

SEEG · epilepsy · seizure onset zone · AUC · conditional null · electrode shaft · spatial sampling · intracranial EEG