Separate-contact confirmation gave lower HFO crossing excess in four of eight sleep SEEG records at a 5% early crossing rate

We tested two-channel high-gamma confirmation in public OpenNeuro ds003498 release 1.1.1. Eight patients supplied one sleep SEEG record each. The two rules used either one shared contact or separate contacts on the same shaft. Each rule had a 5% crossing rate in the first 120 s. Later HFO crossing excess was 14.21 percentage points for shared-contact confirmation and 13.28 for separate-contact confirmation. The mean difference was −0.93 points. The 95% patient bootstrap interval was −4.02 to +2.27 points. Four patients had a lower separate-contact excess, and four had a higher excess. The comparison does not establish a consistent reduction or equivalence. Controlled digital inputs showed a tradeoff. A 100 ms sine added to one shared contact increased crossings by 63.79 points for shared confirmation and 8.55 for separate confirmation. With an alternating input on four contacts, the increases were 62.93 and 52.58 points. The separate-contact rule thus rejected more one-contact input and also lost some distributed-input response under the tested assumptions. Contact distances differed. The source marks were automatic. The data had no intended-command or stimulation outcome labels. We checked all eight fresh raw files and rebuilt all archived power arrays and labels exactly. The package has three figures, source code, complete features, four-contact excerpts, decisions, source hashes, independent checks, and repeat instructions.

intracranial EEG · high-frequency oscillations · signal processing · brain-computer interface · SEEG · threshold crossings