In 24 SEEG patients, 8-s bipolar-pair band-power ranks stayed stable across 10-44 min (HFA rho 0.96) without detectable drift; 2-s windows were less stable

We measured log10 band power (1-150 Hz, 6 bands) in 2966 adjacent same-shaft bipolar SEEG pairs from 24 patients (OpenNeuro ds005624, ses-0). We used 4 windows of 8 s at 10%, 37%, 63% and 90% of each recording. For each patient, band, and window pair, we calculated the Spearman rho across bipolar pairs. The plan was written before we read the signals. Rank agreement was high in all bands. For 70-150 Hz power (HFA), the mean rho between the first and last windows (10-44 min apart) was 0.964 (95% CI 0.951-0.975). Of the HFA top-decile pairs, 76% stayed in the top decile (chance 10%). The primary test did not find a decrease of rho with time separation (HFA slope -0.094 Fisher-z per 10 min, CI -0.241 to 0.057, p 0.24). The CI limits the HFA drift to approximately 0.02 rho per 10 min. When we used only the first 2 s of each window, the rank agreement decreased in all 6 bands (p <= 0.003). The decrease was 0.05-0.07 rho for delta to alpha and 0.01 rho for HFA. Limits: the events files were empty, thus the task, sleep state and seizure times are unknown. Stable non-neural factors (contact position, impedance) can cause part of the rank stability. We used only ses-0 and 8-s windows. The package contains code, byte ranges and SHA-256 of the data blocks, per-pair features, and the results.

SEEG · intracranial EEG · band power · rank stability · bipolar pairs · high-frequency activity · window duration