Selection of contacts in named adjacent pairs with different onset labels reduced mean AUC by 0.130 for a fixed low frequency score in an exploratory test of 34 SEEG patients

We tested whether a fixed low frequency power score could order named adjacent SEEG contacts with different clinical seizure onset labels. We used OpenNeuro ds004100 version 1.1.3 and two 120-second periods per patient. The planned test included 126 pairs in 34 patients. Mean local concordance was 0.508, with a 95 percent patient bootstrap interval of 0.418 to 0.598. Mean all-contact AUC was 0.648 in the same patients. The paired decrease was 0.140, with Holm p 0.00266. A later exploratory control gave AUC 0.518 on contacts in those pairs. Contact selection changed mean AUC by -0.130, with an interval of -0.202 to -0.060. Restricting comparisons on that selected set added -0.010. Most of the observed mean gap came from contact selection. This is an arithmetic result on shared source patients. It does not establish a cause, equivalence to chance, or treatment benefit. The package preserves the features, one raw prefix, source records, code, results, checks, and retrieval instructions.

stereo-EEG · seizure onset zone · local concordance · AUC · contact selection · power spectrum · epilepsy