Absolute and relative beta changes had opposite signs at a 39% weighted rate in three-second SEEG pairs from 21 patients

We tested whether total-power division changes the sign of a beta change in public SEEG. We used OpenNeuro ds006519 version 1.0.0 and three-second before and after pairs from 21 patients. The sample had 35 records and 1,750 retained channel records. Beta was 13 to 30 Hz. The denominator was 1 to 40 Hz. Absolute and relative changes had opposite signs at a 39.2% weighted rate. The 95% patient bootstrap interval was 33.6% to 45.1%. Both magnitudes were at least 1 dB at an 11.9% rate, with interval 7.9% to 16.2%. Each patient had equal weight after shaft and record aggregation. Intervals assume independent patient sampling. The eight specified checks gave rates from 35.4% to 39.2%. Matched controls also had sign disagreement within before and after periods. The result measures a power-definition limit. It does not identify a causal stimulation effect. The package preserves code, paired signals, spectra, patient estimates, resampling draws, and fixed raw retrieval records. All 65 paired signal files were rebuilt from original bytes. Separate calculations checked powers, aggregation, and inference. A second offline run reproduced all 96 numerical files exactly.

stereoelectroencephalography · SEEG · beta power · relative power · spectral power · normalization · Welch method · cortical stimulation · measurement audit