Baseline agreement did not detect better beta frequency matches than peak strength in two-second post-stimulation SEEG windows from 21 people

We tested a short calibration check for beta frequency selection in a brain-computer interface reader. The source was OpenNeuro ds006519 version 1.0.0. We used all 35 SEEG recordings from 21 people. The agreement rule kept channels whose maxima in two baseline windows differed by at most 2 Hz. The strength rule kept the same number of channels with the largest mean baseline residual maximum. Both rules used four baseline seconds. Each rule selected 612 channel occurrences. A match required a later two-second post maximum within 2 Hz of the fitted baseline frequency. Agreement gave 42.86% matches. Strength gave 44.50%. The paired person mean was -1.64 percentage points. Its 95% bootstrap interval was -6.47 to +2.80 points. The exact sign-flip reference gave p = 0.522. All eight fixed sensitivity intervals included zero. Both rules exceeded a within-record frequency reassignment expectation. This test did not detect better matches from agreement. It does not prove equal rules or useful online control. The package preserves three figures, code, compact data, native signal excerpts, source hashes, and reproduction instructions. All raw hashes match the official version tag. A fresh extraction and a separate numerical check passed. This exploratory test reuses a public cohort and a cited spectrum package.

stereoelectroencephalography · SEEG · beta frequency · spectral stability · brain-computer interface · cortical stimulation