Spread sampling gave an uncertain gain in HFO list agreement at 60 seconds of exposure in seven sleep SEEG patients

We tested whether five separated samples improve high frequency oscillation (HFO) list agreement with other sleep SEEG records. Both designs used 60 seconds of exposure. The source was OpenNeuro ds003498 version 1.1.1. The study used 135 records from seven patients with depth electrodes. All integer-second origins received equal weight under circular sampling. Spread agreement was 65.56%. Continuous agreement was 65.05%. The equal-patient gain was 0.52 percentage points. Its 95% patient bootstrap interval was -0.02 to +1.01. The exact sign-flip p value was 0.125. Thus, the data did not show a clear gain. Spread samples were empty less often. Selecting one top pair changed the mean contrast direction. We also prove equal expected agreement under independent, uniform event starts and a fixed separate target. The study measures detector-mark reproducibility. It does not test surgical accuracy. The package preserves source annotations, code, counts, scores, random draws, checksums, and reproduction instructions. All 12 numerical files were reproduced exactly. This is an exploratory study in a reused cohort.

high frequency oscillations · HFO · stereoelectroencephalography · SEEG · epilepsy · sampling design · reproducibility · event detection