Mean clinical seizure onset discrimination by local reference 70–150 Hz power was higher within shafts than across shafts in 20-second samples from eight selected SEEG cases

We tested whether absolute 70–150 Hz power separates clinical seizure onset labels better within an electrode shaft or across shafts. We used public OpenNeuro ds004100, version 1.1.3, with a local neighbor reference. Two recordings each supplied a 20-second sample. Eight of 12 selected cases had both control types. Both comparisons used the same 29 seizure onset centers, with equal case weights and equal positive-contact weights within each case. In run 1, mean discrimination was 0.834 within shafts and 0.626 across shafts. The across-minus-within difference was -0.208. The case bootstrap 95 percent interval was -0.392 to -0.044. Seven of eight differences were negative. The exact two-sided sign test gave p = 0.0703. Run 2 gave means of 0.804 and 0.636 in the same people. These estimates describe short samples from selected cases. They do not establish a population effect or a treatment rule. The package preserves a spatial benchmark, all exclusions, code, source metadata, derived measurements, figures, and an exact raw-data retrieval recipe. A new download reproduced all 24 source power tables and arrays. A fresh offline repeat reproduced all 11 numerical outputs.

stereoelectroencephalography · seizure onset zone · high-frequency power · local reference · spatial sampling · electrode shaft · discrimination