SEEG 4–13 Hz Peak Frequencies Were Moderately Stable Across Two RAM FR1 Sessions

We assessed the stability of 4–13 Hz spectral peak frequency across two sessions in 85 patients performing the RAM free-recall (FR1) task. From archived 60 s spectra, we matched 7,353 bipolar pairs with acceptable data in both sessions and computed peak frequency after removing a 2–40 Hz log-power fit. Across pairs within each patient, the median between-session Spearman correlation was 0.54 (95% CI 0.51–0.59; 93% of 82 patients had r > 0; Holm-adjusted p = 0.00015). Peak height was more stable (median r = 0.68), while peak presence agreement was lower (median Cohen kappa = 0.25). The planned comparison suggested higher between-session than within-session stability, but it compared 60 s sessions with 30 s halves. In the exploratory length-matched comparison, between-session stability was lower (median r = 0.40) than within-session stability (median r = 0.47); the median Fisher-z difference was −0.08 (sign-flip p = 0.0035; bootstrap 95% CI −0.12 to 0.005). Thus, 4–13 Hz peak frequency showed moderate rank stability, while the apparent planned advantage was explained by segment length; any session effect was small and uncertain. Session dates and intervals are unknown. Results are limited by the 60 s segments, coarse 0.5 Hz bins, peak rule, and uncontrolled behavior. The analysis plan was written and hashed before peak results were computed.

stereoelectroencephalography · SEEG · spectral peak frequency · test-retest reliability · RAM free recall · intracranial EEG · theta · alpha