Five second phase controls reduced mean fitted SEEG coincidence score error in nine selected adults with epilepsy

We tested whether local phase controls reduce the error in a simultaneous extreme sample score. We used 36 windows from rest and film records in nine selected adults with epilepsy in OpenNeuro ds003688, version 1.0.7. Each method used 199 reference draws. Five second controls reduced mean absolute score error from 1.945 to 1.335 per 1000 pair samples. The mean gain was 0.611, or 31.4 percent. The participant bootstrap 95 percent interval was 0.351 to 0.896. The exact sign-flip probability was 0.00390625 under conditional sign symmetry. All nine complete estimates were positive. The smallest estimate changed sign when the draws were split in half. Gaussian signals with shared gain changes showed a possible mechanism. The controls use constraints fitted to the same windows. The result does not identify nonlinear brain coupling or prediction on new data. This is a method follow-up on the earlier cohort. The package contains all 18 compact inputs, code, every reference score, source hashes, figures, and repeat instructions. A complete repeat matched all results exactly. Five native records passed fresh reconstruction. A larger raw recheck had download timeouts.

stereoelectroencephalography · phase surrogates · Fourier phase randomization · cross spectra · simultaneous extremes · surrogate data · epilepsy · nonstationarity