In 42 patients, age was not significantly associated with NREM sigma peak frequency in normal-contact SEEG pairs after cohort adjustment

Question: scalp EEG shows that sleep spindle frequency increases from childhood to adulthood. Is this also true for local SEEG signals? Data: OpenNeuro ds007118, ds007119, ds007120 (snapshot 1.0.0), first run of each SEEG subject, 30-120 s, read by HTTP byte range with SHA-256 per range. Bipolar pairs of adjacent contacts with the clinical label 'normal' (rules and reader from AutoSci 5c22e5d3): 54 patients, 711 pairs. We wrote the plan before we read any signal. Method: Welch PSD (0.25 Hz bins), a log-log aperiodic line fitted at 3-8 and 17-30 Hz, a sigma peak if the 9-16 Hz residual is >= 0.10 log10 units and not at an edge. 541 pairs had a peak. Primary test: patient median peak frequency (42 patients with >= 3 peak pairs) against age, Spearman rho partial on cohort, 20000 within-cohort permutations, 5000 within-cohort bootstrap samples. Result: partial rho +0.19 (95% CI -0.17 to +0.54, p = 0.28; raw rho +0.06), within-cohort slope +0.027 Hz per year. The hypothesis is not supported. Peak prevalence (partial rho +0.02) and peak height (+0.05) also did not change with age (Holm p = 1.0). All sensitivity analyses (no resected contacts, age >= 4 y, thresholds 0.05 and 0.15, leave one cohort out) gave positive rho from +0.09 to +0.24, all p > 0.2. Pooled adults had lower peak frequency (12.2 Hz) than children (13.3 Hz). An exploratory per-cohort table shows that this comes from differences between cohorts: the youngest cohort had the fastest peaks, but inside the two larger cohorts rho was positive (+0.30, +0.28). Limits: 42 patients give 80% power only for |rho| of about 0.42 or more; drug-resistant epilepsy; one 90-s segment; no N2/N3 separation; contact region not controlled; spectral peak, not discrete spindle events.

NREM sleep · sleep spindles · sigma peak frequency · SEEG · intracranial EEG · age · epilepsy · cohort adjustment