In 256 adult RAM patients, older age was associated with shorter SEEG aperiodic-knee timescales, but not independently of fitted exponent

We tested the pre-registered prediction that the stereo-electroencephalography (SEEG) aperiodic-knee timescale is shorter at older age in 256 adults (18–64 years) with epilepsy from the RAM study. The analysis used archived 2–40 Hz per-pair knee fits from 60-second segments at the start of a memory task in OpenNeuro ds004789 (FR1) and ds004809 (catFR1); it did not make new fits. Subject-level timescale was the median across bipolar pairs. In a partial Spearman analysis adjusted for dataset, sex, site, sampling-rate class, and pair count, age was associated with shorter median log10 timescale (rho = −0.269, 95% CI −0.382 to −0.148, permutation p = 0.0001; n = 256). The association remained in sensitivity analyses, including analyses without grid-edge fits, lateral temporal pairs, location adjustment, and each dataset separately. The fitted exponent also increased with age (rho = +0.326, 95% CI +0.209 to +0.434, Holm p = 0.0005). In a post hoc analysis adjusted for exponent, the timescale association was not present (rho = −0.047, 95% CI −0.172 to +0.080, p = 0.45). The results therefore show an age-related change in the fitted aperiodic spectral shape, but do not establish a timescale change independent of exponent. The sample included epilepsy patients only, one task segment per patient, no seizure-onset-zone exclusion, and no children or adults older than 64 years. The cross-sectional results do not show within-person change.

stereo-electroencephalography · aperiodic activity · knee frequency · timescale · aging · RAM study · partial Spearman correlation