Sex explains little of patient-level SEEG spectral variation in 255 RAM patients; peak-frequency evidence is inconclusive

We tested whether sex explains patient-level variation in the 4–13 Hz SEEG peak and five other bipolar spectral features in 255 RAM epilepsy patients (131 women, 124 men). We analyzed archived tables from ds004789 and ds004809, adjusted features for age, sampling rate, dataset and site, and used 10,000 within-site sex permutations and 2,000 patient bootstraps. The peak was 0.29 Hz higher in women (95% CI 0.04–0.55; p=0.020; Holm p=0.12). No test survived Holm correction, and the peak interval does not exclude a difference of 0.3 Hz; this result is inconclusive. Relative theta, relative beta and aperiodic exponent had nominal differences, each no larger than 0.3 SD. Relative alpha and broadband power showed no difference; for broadband power the estimate was +0.03 SD (95% CI −0.23 to +0.26). In a replication cohort of 44 new patients, the peak difference was +0.09 Hz (95% CI −0.60 to +0.94; p=0.82). This replication is too imprecise to confirm or reject the discovery estimate. The cohorts were clinical epilepsy patients; sex came from binary F/M participant metadata, and disease, medication and implant strategy could confound the results. The replication used subject-level location covariates, and only 44 of 57 new people had coordinates.

SEEG · spectral features · sex differences · epilepsy · peak frequency · replication