Peak amplitude scores had a higher mean AUC than peak isolation scores for late peak time agreement in five SEEG patients

We tested whether a peak isolation score from five pulses ranked late peak time agreement with five unused pulses better than a peak amplitude score. The source was public OpenNeuro ds004457, version 1.0.2. The sample had 1,081 stimulation site and receiver pairs in five patients with SEEG implants. SEEG means stereoelectroencephalography. We searched absolute peaks from 50 to 250 ms after a pulse. We subtracted a before-pulse mean and used an 11-bin box mean at 512 Hz. Agreement meant a peak time gap of at most 20 ms. Both scores used the same early pulse variation scale. Mean patient AUC was 0.630 for isolation and 0.689 for amplitude. AUC means area under the receiver operating characteristic curve. The paired difference was -0.059, with empirical limits from -0.091 to -0.023. These limits describe five-patient resamples; they are not a calibrated population interval. Mean agreement in the highest quarter was 0.498 for isolation and 0.595 for amplitude. All nine fixed after-pulse settings had a negative mean AUC difference for isolation minus amplitude. A check added after analysis removed one pair with a constant pulse and gave the same direction. The margin proof needs a bound on waveform change. The tested pulse variation scale does not provide that bound. Neither score gives a timing guarantee or detects a neural response. The package preserves the local fixed plan, the marked addition, data, full results, source locks, code, and repeat instructions. A fresh raw rebuild matched every saved array. A second analysis gave identical bytes for all 17 numerical result files.

stereoelectroencephalography · peak timing · signal processing · AUC · electrical stimulation