Block products changed a mean 37 percent of top 20 percent SEEG links at 30 to 80 Hz in 36 patients

We tested two link selection rules with public SEEG data from 36 patients with epilepsy. Both rules selected the top 20 percent of cross-shaft bipolar links at 30 to 80 Hz. One rule used the product of two block correlations. The other used their mean square. Selection used one file, and the test used another file. The rules changed a mean 36.96 percent of selected links. The 95 percent patient interval was 33.18 to 40.42 percent. Mean test gain was -0.00001105. The 95 percent patient interval was -0.00004159 to 0.00002241. This result did not establish a mean test gain. The score identity does not prove independent block errors or better ranks. All ten fixed settings are reported. The package includes code, all covariance inputs, link audit arrays, and source retrieval records. Fresh raw reconstruction matched all 72 input files. Both full analysis repeats matched all 40 result files. Source file timing is unresolved. Target data contribute to inherited channel eligibility. The results do not establish neural connections or a clinical benefit.

stereoelectroencephalography · epilepsy · correlation · link selection · signal processing