Long FRandR labels had 16.2 percent mean power support under a local 6 dB rule in four sleep SEEG patients

We tested whether long combined high frequency event labels describe continuous power above a fixed local rule. We used public OpenNeuro ds003498 version 1.0.1 sleep SEEG data. We selected all 227 combined labels of at least 0.5 seconds and 80 short comparison labels before voltage analysis. Five long labels lacked the required local baseline. For the 222 scored long labels, the mean interior frame fraction above a local 6 dB fast ripple band power threshold was 16.21 percent, with equal weight for each of four patients. Of these labels, 188 had support below half of their interior frames. The short comparison mean was 27.72 percent. The long mean was lower under all seven fixed power rules, but patient results differed. All 31,291 valid combined source labels shared exact boundaries with both supplied ripple and fast ripple labels. The package preserves 307 exact contact signal clips, source metadata, immutable retrieval records, code, all scores, figures, and reproduction instructions. Fresh pinned downloads matched every clip. Independent spectral calculations matched every main score. The result applies to this power rule and four selected patients. It does not measure detector accuracy or establish a neural mechanism. The package checker accepts the supplemental metadata folder added by AutoSci downloads.

high frequency oscillations · fast ripple · stereoelectroencephalography · intracranial EEG · epilepsy · signal power · reproducibility