1 August 2020 in Epilepsia
22 September 2022 in Epilepsia
Objective: Epilepsy is characterized by spontaneous seizures that recur at unexpected times. Nonetheless, using years‐long electroencephalographic (EEG) recordings, we previously found that patient‐reported seizures consistently occur when interictal epileptiform activity (IEA) cyclically builds up over days. This multidien (multiday) interictal–ictal relationship, which is shared across patients, may bear phasic information for forecasting seizures, even if individual patterns of seizure timing are unknown. To test this rigorously in a large retrospective dataset, we pretrained algorithms on data recorded from a group of patients, and forecasted seizures in other, previously unseen patients.Methods: We used retrospective long‐term data from participants (N = 159) in the RNS System clinical trials, including intracranial EEG recordings (icEEG), and from two participants in the UNEEG Medical clinical trial of a subscalp EEG system (sqEEG). Based on IEA detections, we extracted instantaneous multidien phases and trained generalized linear models (GLMs) and recurrent neural networks (RNNs) to forecast the probability of seizure occurrence at a 24‐h horizon.Results: With GLMs and RNNs, seizures could be forecasted above chance in 79% and 81% of previously unseen subjects with a median discrimination of area under the curve (AUC) = .70 and .69 and median Brier skill score (BSS) = .07 and .08. In direct comparison, individualized models had similar median performance (AUC = .67, BSS = .08), but for fewer subjects (60%). Moreover, calibration of pretrained models could be maintained to accommodate different seizure rates across subjects.Significance: Our findings suggest that seizure forecasting based on multidien cycles of IEA can generalize across patients, and may drastically reduce the amount of data needed to issue forecasts for individuals who recently started collecting chronic EEG data. In addition, we show that this generalization is independent of the method used to record seizures (patient‐reported vs. electrographic) or IEA (icEEG vs. sqEEG).
31 January 2020 in Epilepsia Open
Objective: Neurostimulation devices that deliver electrical impulses to the nervous system are widely used to treat seizures in patients with medically refractory epilepsy, but the effects of these therapies on sleep are incompletely understood. Vagus nerve stimulation can contribute to obstructive sleep apnea, and thalamic deep brain stimulation can cause sleep disruption. A device for brain‐responsive neurostimulation (RNS®System, NeuroPace, Inc) is well tolerated in clinical trials, but potential effects on sleep are unknown.Methods: Six adults with medically refractory focal epilepsy treated for at least six months with the RNS System underwent a single night of polysomnography (PSG). RNS System lead locations included mesial temporal and neocortical targets. Sleep stages and arousals were scored according to standard guidelines. Stimulations delivered by the RNS System in response to detections of epileptiform activity were identified by artifacts on scalp electroencephalography.Results: One subject was excluded for technical reasons related to unreliable identification of stimulation artifact on EEG during PSG. In the remaining five subjects, PSG showed fragmented sleep with frequent arousals. Arousal histograms aligned to stimulations revealed a significant peak in arousals just before stimulation. In one of these subjects, the arousal peak began before stimulation and extended ~1 seconds after stimulation. A peak in arousals occurring only after stimulation was not observed.Significance: In this small cohort of patients, brain‐responsive neurostimulation does not appear to disrupt sleep. If confirmed in larger studies, this could represent a potential clinical advantage of brain‐responsive neurostimulation over other neurostimulation modalities.
28 September 2016 in bioRxiv
Theta oscillations play a critical role in learning and memory by coordinating the spiking activity of neuronal ensembles via mechanisms such as spike timing dependent plasticity 1–7 . This rhythm is present in rodents where it is continuously evident during movement at frequencies within 6-12Hz 8,9 . In humans, however, the presence of continuous theta rhythm has been elusive; indeed, a functionally similar theta is thought to occur at lower frequency ranges (3-7Hz) and in shorter bouts 10–12 . This lower frequency theta rhythm is observed during a variety of behaviors, including virtual navigation, but has never been tested during real world ambulatory movement . Here we examined the oscillatory properties of theta within the human medial temporal lobe (MTL) in freely moving human subjects chronically implanted with the clinical NeuroPace RNS ® responsive neurostimulator device, capable of wireless recordings of continuous intracranial deep brain electroencephalographic (iEEG) activity. MTL iEEG recordings, together with sub-millimeter position tracking, revealed the presence of high frequency theta oscillations (6-12Hz) during ambulation. The prevalence of these oscillations was increased during fast movement compared to slow movement. These theta bouts, although occurring more frequently, were not significantly different in durations during fast versus slow movements. In a rare opportunity to study one subject with congenital blindness, we found that both the prevalence and duration of theta bouts were much greater than those in sighted subjects. Our results suggest that higher frequency theta indeed exists in humans during movement providing critical support for conserved neurobiological mechanisms for spatial navigation. The precise link between this pattern and its behavioral correlates will be an exciting area for future studies given this novel methodology for simultaneous motion capture and long term chronic recordings from deep brain targets during ambulatory human behavior.