Interictal Epileptiform Discharges Show Multidien Cycles in Ultra Long-term Subcutaneous EEG
Abstract number :
1.167
Submission category :
3. Neurophysiology / 3C. Other Clinical EEG
Year :
2022
Submission ID :
2204278
Source :
www.aesnet.org
Presentation date :
12/3/2022 12:00:00 PM
Published date :
Nov 22, 2022, 05:24 AM
Authors :
Asbjoern Helge, MSc – UNEEG medical; Pedro Viana, MD – King's College London; Mark Richardson, PhD, MD – King's College London; Troels Kjaer, PhD, MD – UNEEG medical; Jonas Duun-Henriksen, PhD, MSc – UNEEG medical
Rationale: Research in the recent decade have revealed occurrences of cyclical multidien patterns of seizures and interictal epileptiform discharges (IED) in people with epilepsy (PWE) (Karoly et al., 2021). These findings have been made possible by intracranial recordings with a duration of up to several years, but similar results have not yet been demonstrated using less invasive electroencephalography (EEG) recording methods. However, novel minimally invasive devices for obtaining ultra long-term subcutaneous EEG have made it possible to investigate this multidien pattern without the cost of major surgery. This may enable faster medication adjustment or seizure forecasting. In this work, we set out to find multidien cycles of IEDs in ultra long-term subcutaneous EEG using a periodogram method developed for unevenly sampled data.
Methods: A subset of EEG recordings was selected from a dataset consisting of data from nine PWE (Weisdorf et al., 2019). The PWE were monitored with the 24/7 EEG™ SubQ solution for approximately three months. Data from PWE were included if a neurologist had previously observed IEDs in the EEG and if the PWE had had a device adherence above 50%. An automatic algorithm was implemented to locate IEDs. For all PWE, a Lomb-Scargle periodogram was computed based on the half-hour IED rate with a minimum period length of 4 days and a maximum period length set to one fifth of the recording length. The Lomb-Scargle periodogram was chosen as this method accommodates gaps in the data (VanderPlas, 2018). The probability that a signal with no periodic component would lead to a peak of this magnitude was computed for the most prominent peak in each periodogram. If this probability was below 0.01, a multidien cycle was found with a cycle length corresponding to the most prominent peak.
Results: Data from five PWE were included in the analysis with adherence rates between 63% and 91%. Four of them had multidien IED cycles of approximately 5, 9, 12, and 15 days with probabilities of < 0.001, < 0.1e-5, < 0.1e-5, and < 0.1e-10.
Neurophysiology