Seizure forecasting based on AI-supported analysis of multidien and circadian cycles in EEG and non-EEG long-term datasets

Abstract Long-term datasets in epilepsy encompassing weeks to months of continuous physiological signal recordings along with novel data analysis techniques have recently advanced the understanding of epilepsy in several aspects. Patterns of seizures, interictal discharges, and autonomous nervous system activity were observed to often exhibit long, multidien cycles that are often correlated with each other. These observations have provided the basis for new approaches to forecast seizure risk from electroencephalographic (EEG) and non-EEG data.