Energy and spectrum controls for 1 s blocks kept a mean seizure onset label AUC of 0.648 in 1–45 Hz SEEG from 30 patients

We tested an increment concentration score in public SEEG from OpenNeuro ds004100 version 1.1.3. The fixed sample had 30 patients, two interictal clips per patient, and 537 bipolar pairs in 69 electrode name groups. The signals had a 1–45 Hz Fourier projection. The analysis used ordinary increments from a 58 s interval. AUC is the rank score for seizure onset labels within each name group. Observed mean AUC was 0.7030. Random phase controls that kept energy, spectrum, and mean in each 1 s block had mean AUC 0.6479. Controls for the complete interval had mean AUC 0.5028. The primary observed minus local control difference had mean 0.0551 and descriptive patient bootstrap interval 0.0122–0.1022. The difference was positive in 17 patients and negative in 13 patients. The planned exact two-sided sign test gave p = 0.5847. An exact decomposition separates scale variation between blocks from a weighted term within blocks. A Gaussian scale mixture showed that scale variation can create a global control gap. All 60 public source prefixes were retrieved again. All rebuilt signal arrays matched. A fresh-process replay reproduced all 66 analysis result files. The package preserves signals, source records, control scores, results, code, and instructions. This secondary study does not isolate a neural mechanism or test clinical benefit. The inherited numeric coordinate rule has unresolved units.

stereo-EEG · seizure onset labels · increment concentration · Fourier phase controls · signal processing · patient bootstrap · OpenNeuro ds004100