A test did not detect a rise in excess squared residual correlation between shafts at 100 to 500 ms after stimulation in seven selected SEEG patients under patient independence and null sign symmetry

We tested whether excess squared residual waveform correlation between different SEEG shafts rises after single-pulse stimulation. The public source was OpenNeuro ds004696 version 1.0.1. The fixed selection had seven patients and 80 stimulation groups, with ten good exported pulses per group. Adjacent same-block contact differences gave 4,218 edges between distinct named shafts. We removed each pulse time mean and the mean waveform in two blocks of five pulses. The score was matching-pulse squared residual correlation minus its exact mean under within-block relative pulse permutations. This benchmark retains each pulse waveform and its temporal correlation. Patients had equal weight. The post window was 100 to 500 ms. The mean post-minus-pre change was +0.00111. The descriptive 95 percent patient interval was -0.00345 to +0.00553. The exact two-sided sign-flip p value was 0.6875 under patient independence and null sign symmetry. Five patients had positive changes. Planned descriptive checks had mixed directions and intervals that included zero. The test did not detect a rise. It does not establish a zero effect or physiological independence. The metric can miss shared inputs with different shapes or time delays. All seven freshly retrieved source files rebuilt the saved voltage arrays exactly. Independent calculations and a new process reproduced the result. The package preserves selected voltages, code, a proof of the benchmark, source versions and hashes, results, figures, and repeat instructions.

SEEG · electrical stimulation · residual correlation · waveform analysis · permutation benchmark · epilepsy