The propensity for seizures to follow circadian and multiday (i.e., weekly, monthly, or seasonal) rhythms has been documented for centuries. More recent findings from chronically recorded EEG in both human and animal studies have further elucidated the existence of multiday rhythms governing seizure timing and rates of epileptic activity. These multiday epileptic cycles are found to be prevalent in a number of studies, including from implantable devices , electronic seizure diaries, and wearable monitoring. It is becoming clear that multiday seizure cycles are important phenomena in epilepsy, with many implications for seizure management and more broadly in interpreting research studies and clinical trials.
The integration of knowledge of seizure cycles into clinical practise is in an early phase, with some theoretical studies demonstrating the utility of seizure risk forecasts or scheduling diagnostic testing based on multiday cycles. However, several barriers remain before seizure cycles can be widely adopted in clinical management. One barrier is the knowledge gap between data scientists and clinicians. Cycles are measured and described using circular statistics and frequency analysis, making the explanation of the presence and strength of cycles technically complex. It is yet to be determined at what strength a cycle can be deemed to be clinically relevant for making treatment or monitoring decisions. Much progress has been made converging the clinical and engineering aspects of epilepsy, however seizure cycles present new concepts in seizures and epilepsy that need to be clearly interpretable by clinicians. This review aims to bridge several gaps that commonly arise between data science and clinical practise.
Philippa J. Karoly, Dean R. Freestone, Dominique Eden, Rachel E. Stirling, Lyra Li, Pedro F. Vianna, Matias I. Maturana, Wendyl J. D’Souza, Mark J. Cook, Mark P. Richardson, Benjamin H. Brinkmann and Ewan S. Nurse
Published on 04 August 2021
Frontiers in Neurology