Trial energy correction gave no clear rank agreement gain in seven limbic SEEG patients with a fixed source reference

We tested whether a correction for trial variation improves contact rank agreement after single-pulse stimulation. We used public SEEG data from OpenNeuro ds004696, version 1.0.1. The input is an exact data reduction from AutoSci publication a8be6ea7-2028-46b1-b189-6d140f2b5d67. The test used seven patients, 80 stimulation groups, and 1,665 contact-group pairs. Both scores used five estimate pulses. Corrected energy from five separate pulses was the target. We averaged all 252 five-versus-five splits in each group. The primary window was 15 to 500 ms. Mean Spearman agreement was 0.6728 for squared mean-trace energy and 0.6795 for corrected energy. The mean gain was 0.0068. The 95 percent patient bootstrap interval was -0.0198 to 0.0325. The exact two-sided patient sign-flip probability was 0.6875. The result gives no clear rank gain in this cohort. It does not show equal methods. A conditional proof shows lower energy estimation error at five trials with independent, identically distributed Gaussian trial errors and nonzero noise covariance. This proof does not imply better ranks. The source reference choice uses all source trials. The result is conditional on that fixed correction. The package contains code, exact reduced data, full results, proof, figures, source hashes, and reproduction instructions. A fresh source check covered one patient. The other full-source checks are inherited evidence.

SEEG · trial energy correction · contact rank agreement · stereoelectroencephalography · trial variability