Aperiodic Neural Dynamics in Stereo-eeg Track Seizure Activity, Region, and Circadian Phase in Focal Epilepsy

Introduction: Aperiodic (1/f) activity, represented by the slope of the power spectrum, has gained attention as a biomarker of neural excitation–inhibition balance. While prior work has emphasized oscillatory dynamics in epilepsy, emerging evidence suggests aperiodic activity may reveal important features of epileptogenic networks. However, how aperiodic exponent tracks seizure dynamics remains poorly understood. The objective of this study was to characterize spatial and temporal patterns of aperiodic activity in focal epilepsy, including changes across ictal states and circadian rhythms.

Methods: We analyzed stereo EEG recordings from seven patients with drug-resistant epilepsy with mesial temporal seizure onset zones who received responsive neurostimulation (RNS). One-minute epochs were sampled hourly throughout intracranial monitoring. Recordings were segmented into 30-second ictal and peri-ictal epochs from hippocampus and supplementary motor area channels. Peri-ictal epochs were related to seizure activity (seizures within 30 minutes of that hour). Additionally, one-minute epochs were analyzed in 15-minute increments ±60 minutes relative to seizure onset, classified as preictal, ictal, and postictal. Power spectra (1–55 Hz) were parameterized with SpecParam to obtain aperiodic exponent. Mixed-effects regression examined effects of region, laterality, and peri-ictal episodes. Circadian variation in exponent values was evaluated via cosinor models.

Results: Seven patients (mean age 38.7, 5 females) with average monitoring duration of 5.78 days yielded 971 epochs. Mean seizure reduction after RNS was 83.4% at follow-up (mean 8 months). Temporal fluctuations in the exponent were observed throughout monitoring. Relative to the SMA, the hippocampus showed a lower exponent (0.13, 95% CI −0.156 to −0.111, p< 0.001) while left hemisphere showed higher exponent (0.14, 95% CI 0.116–0.154, p< 0.001). Compared to interictal periods, peri-ictal epochs were associated with an increase in the exponent (0.07, 95% CI 0.034–0.113, p< 0.001). Circadian rhythmicity was present in most patients, with individual differences in peak timing over 24 hours. Correlations between seizure reduction and day–night exponent differences were not significant. Aperiodic exponent changes from ±60 minutes of seizure onset revealed region-specific patterns. SMA remained stable across preictal to postictal phases, while hippocampus showed decreased exponent from preictal to ictal and subsequent recovery. Individual trajectories showed marked heterogeneity, with some showing ictal suppression while others exhibited delayed, minimal, or multiphasic changes, suggesting patient-specific dynamics.

Conclusion: Aperiodic neural activity varies systematically with seizure state, anatomical region, and circadian phase. These findings support the aperiodic exponent as a potential biomarker of epileptogenic networks and time-varying cortical state, with translational relevance for neuromodulation timing and targeting.