Epileptiform discharges (EDs), including ictal and interictal activity, exhibit structured temporal patterns across hours, days, and months. The physiological mechanisms underlying these rhythms, however, remain poorly understood.
To investigate ultradian and circadian variation, we analyzed 24-hour EEG recordings from 107 individuals with idiopathic generalized epilepsy and identified two subgroups with distinct ED distributions. To explore potential drivers, we developed a dynamic brain network model describing transitions between background and seizure-like states. A time-dependent forcing term captured physiological modulation of network excitability, with parameters informed by EEG-derived ED distributions, sleep stages, and hormone dynamics from blood samples. Sleep accounted for most variability in one subgroup, while hormone fluctuations explained the majority in the other. Integrating both measures improved model fit in the first subgroup, suggesting distinct physiological contributions to ED rhythms.
Building on these findings, we conducted a pilot study in which continuous EEG and cortisol were recorded simultaneously over 24 hours in people with stress-sensitive epilepsy. EEG was recorded using a wearable headset, while cortisol was sampled using U-RHYTHM, a minimally invasive, portable hormone-sampling device. Together, these complementary modalities provide a unique, high-resolution multimodal dataset capturing brain activity and physiological hormone dynamics in real-world conditions. Incorporating these measures into dynamic network models will allow us to understand how sleep and hormone fluctuations jointly shape ED likelihood.
By combining computational modeling with multimodal physiological data, this approach advances understanding of the drivers of epileptiform activity and provides a framework for mechanistic studies that could optimize clinical management in epilepsy.