Computational modeling of the fast-to-slow EEG rhythm transition across seizure evolution and postictal recovery

Epileptic electroencephalography (EEG) can exhibit progressive rhythm slowing across seizure evolution and postictal recovery. How fast ictal activity and slower recovery-related processes generate this fast-to-slow rhythm transition within recurrent seizure cycles remains poorly understood. This study aimed to reproduce and explain this spectral evolution using a dynamical model. To provide an empirical reference, we characterized spectral evolution in 33 seizures from 11 patients using time-frequency analysis. Within-patient analysis showed ictal slowing followed by retained relative low-frequency weighting during early postictal recovery, whereas postictal trajectories varied across patients. We developed a Dual-Component Epileptic EEG (DCE) model comprising two coupled subsystems: one generates fast ictal discharges and spontaneous interictal--ictal transitions; the other governs transitions among fast spikes, spike--wave discharges, and slow waves. The model qualitatively reproduced the fast-to-slow rhythm transition and delineated interictal, ictal, and postictal phases. Three parameter families regulated distinct transition features: timescale ratios set spectral span, filtering rates governed temporal organization, and connectivity coefficients shaped oscillatory-regime composition by shifting bifurcation thresholds. In simulations of two coupled DCE nodes, coupling strength controlled the extent of recruited rhythm slowing, and recruited events exhibited longer model-defined postictal durations than spontaneous events. By linking recurrent interictal--ictal switching with spectral evolution, the DCE model provides a unified dynamical account of fast-to-slow rhythm transitions and their variability. Its model-defined ictal--postictal boundary may support patient-specific estimation of electrographic seizure offset. Inter-node differences in rhythm slowing and postictal duration may serve as candidate dynamical signatures for distinguishing driving from recruited nodes in epileptic networks.