Postgraduate Researcher The University of Melbourne Templestowe, Victoria, Australia
Rationale: It is now well established that seizure occurrence is modulated by circadian and multi-day rhythms, with individual periodicities around daily, weekly, and monthly time scales (Karoly et al 2016, Baud et al 2018). These cycles are prevalent within the population, with at least 80% of people showing circadian seizure cycles and more than 50% of people showing multi-day seizure cycles (Karoly et al 2018). Despite the increasing awareness of cycles in epilepsy, the underlying causes remain unknown, although environmental and physiological factors have been hypothesized to play a role.We report on preliminary results from the ‘Tracking Seizure Cycles’ study, a two-year, prospective cohort study to measure cycles in people with epilepsy via wearables and seizure diaries. Methods: Tracking Seizure Cycles was approved by the St Vincent’s Hospital Human Research Ethics Committee (HREC 009.19) with the first enrollment in August 2019. Participants wore a smartwatch and manually reported seizures in a mobile diary app. The smartwatch continuously measured participants’ heart rates (via photoplethysmograph), estimated sleep stage (wake, REM, stage 2, stage 3), and step count (Figure 1).There are currently 33 participants (13 male) with a cumulative total of 68,541 hours of continuous heart rate recorded and over 3,000 nights of sleep scoring. The total diary duration across participants was 25 years. Participant diaries included 1,558 seizures with 794 seizures reported during the wearable monitoring period.Heart rate, sleep and step count data were analyzed relative to seizure occurrence. Periodic behavior was examined at fast (circadian) and slow (multi-day, about-weekly and about-monthly) scales using a wavelet transform to detect significant cycles at different periodicities. Signals were then bandpass filtered to extract significant cycles and quantify the relationship between cycle phase and seizure onset. Phases where seizures occurred were plotted on circular histograms (Figure 2) and the statistical significance (p < 0.05) of phase-locking was determined by the Rayleigh test. Results: There were 19 out of 33 eligible participants with at least two months of wearables data (range 2 to 8.5 months). Mean adherence for wearable devices was 81%. Analysis of resting heart rate showed eighteen people (95%) had significant circadian cycles, 19 people (100%) had significant about-weekly cycles and six people (32%) had significant about-monthly cycles. Eight out of 19 had more than 20 diary-reported seizures during recording and qualified for seizure phase analysis. Seven of eight (87%) had seizures significantly phase-locked onto at least one of their heart rate cycles. Conclusions: Circadian and multi-day cycles are prevalent in resting heart rate and correspond to seizure occurrence. These novel findings suggest the brain-heart interaction modulates seizure risk directly, or via a secondary physiological mechanism regulating both heart rate and epileptic cycles. The results of this study represent an important step towards developing personalized, non-invasive seizure forecasts. Funding: Please list any funding that was received in support of this abstract.: This project received funding from the Australian Government National Health and Medical Research Council (NHMRC Investigator Grant 1178220) This project was also supported by the Epilepsy Foundation of America’s Epilepsy Innovation Institute “My Seizure Gauge” grant.