In eight epilepsy patients, distant SEEG channels shared little single-trial CCEP variation and a sentinel predicted less than 1%

We asked whether trial-to-trial variation in early (10–50 ms) single-pulse cortico-cortical evoked potential (CCEP) amplitude is shared across distant stereoelectroencephalography (SEEG) channels, as a distant sentinel channel might help a closed-loop stimulator estimate responses elsewhere. We analyzed an existing single-trial table for 8 patients in OpenNeuro STReEF ds005448 v1.0.0, using channel pairs on different shafts at least 20 mm apart. For responder pairs, mean correlation was 0.049 with the first pulse removed and 0.067 with all pulses; corresponding sham means were −0.008 and 0.017 (exact sign-flip p = 0.047 and 0.086). Pairs without clear responses also showed a small real excess (about 0.01). A sentinel based on other distant responders explained less than 1% of target variance (median r² = 0.004) and did not outperform sham (p = 0.31 and 0.87). The results indicate that distant single-trial CCEP variation was mostly local or noise in this clinical sample; they do not support a useful distant sentinel. The source table has a documented stale responder-ratio value in sham rows. This analysis used the ratio only from real rows. Limitations include 8 patients, only 4–5 pulses per cell, a derived table that was not rechecked against raw signals, and no correction for multiple tests.

cortico-cortical evoked potentials · stereoelectroencephalography · single-trial variability · intracranial stimulation · closed-loop brain-computer interface · sentinel channel · trial-to-trial correlation