In seven patients, the polarity-invariant SEEG response did not decrease more than the polarity-flipping component during 43 Hz trains (p = 0.53; undetermined)

This package applies a pre-registered analysis to 12 alternating-polarity, 43 Hz bipolar SEEG stimulation trains from seven patients in OpenNeuro ds006519 version 1.0.0. It tests whether the polarity-invariant pulse-locked component S decreases from the first to last second more than the polarity-flipping component D. The primary contrast was +0.054 (3/7 patient values negative; exact two-sided sign-flip p = 0.53), so the pre-registered decision was undetermined. S was near the noise floor on most channels. Among channels selected for S_first > 3 times noise, S decreased in all seven patients (mean log ratio -0.224, p = 0.0156), but selection on a high first value biases this ratio downward; it is not a valid adaptation test. Polarity labels were reliable in six of 12 trains. A post-hoc strict-alternation analysis gave a primary contrast of +0.037 (p = 0.59); analysis of only trains with alt_frac >= 0.9 included three patients and gave contrast +0.000 (p = 1.0). The exact test with seven patients cannot reach p < 0.0156. S and D may mix neural response and artifact because of nonlinear artifact, amplifier recovery, and polarity-dependent neural responses. The package preserves methods, analysis code, input SHA-256 provenance, result tables, and reproduction instructions.

stereo-electroencephalography · SEEG · electrical stimulation · alternating polarity · neural adaptation · signal decomposition · artifact · sign-flip test · OpenNeuro ds006519