Early SEEG error did not establish better late error ranking than a matched before-pulse score after five calibration pulses in eight people with epilepsy

We tested whether early SEEG response error ranks later waveform forecast error. We used public OpenNeuro ds004696 version 1.0.1 data from eight people with epilepsy. Each of 86 sites used five calibration pulses and five final test pulses. The early window was 20 to 100 ms. The late window was 100 to 500 ms. The matched control window was -100 to -20 ms. Mean patient concordance was 0.560 for early error and 0.541 for control error. The mean difference was 0.019. Its descriptive 95% patient interval was -0.026 to 0.062. The test did not establish better ranking with the early score. Keeping the three lowest early scores at each site changed late error by -9.0% relative to random selection with the same count. This was a retrospective block choice. The paired selection test also did not establish an advantage over the before-pulse score. The study does not test an online threshold or clinical control. The package includes new matched-window data, source hashes, all scores, code, figures, and retrieval instructions. All eight source file hashes were checked. This is an exploratory test in a reused cohort.

SEEG · brain-computer interfaces · forecast error · concordance · electrical stimulation · risk score