Seizure triggers identified postictally using a smart watch reporting system.

Persons with epilepsy (PWE) often report that seizure triggers can influence the occurrence and timing of seizures. Some previous studies of seizure triggers have relied on retrospective daily seizure diaries or surveys pertaining to all past seizures, recent and/or remote, in respondents. To assess the characteristics of seizure triggers at the granularity of individual seizures, we used a seizure-tracking app, called EpiWatch, on a smart watch system (Apple Watch and iPhone) in a national study of PWE. Participants tracked seizures during a 16-month study period using the EpiWatch app. Seizure tracking was initiated during a pre-ictal state or as the seizure was occurring and included collection of biosensor data, responsiveness testing, and completion of an immediate post-seizure survey. The survey evaluated seizure types, auras or warning symptoms, loss of awareness, use of rescue medication, and seizure triggers for each tracked seizure. Two hundred and thirty four participants tracked 2493 seizures. Ninety six participants reported triggers in 650 seizures: stress (65.8%), lack of sleep (30.5%), menstrual cycle (19.7%), and overexertion (18%) were the most common. Participants often reported having multiple combined triggers, frequent stress with lack of sleep, overexertion, or menses. Participants who reported triggers were more likely to be taking 3 or more anti-seizure medications compared to participants who did not report triggers. Participants were able to interact with the app and use mobile technology in this national study to record seizures and report common seizure triggers. These findings demonstrate the promise of longitudinal, self-reported data to improve our understanding of epilepsy and its related comorbidities.

Keyword Spotting Using Human Electrocorticographic Recordings.

Neural keyword spotting could form the basis of a speech brain-computer-interface for menu-navigation if it can be done with low latency and high specificity comparable to the "wake-word" functionality of modern voice-activated AI assistant technologies. This study investigated neural keyword spotting using motor representations of speech via invasively-recorded electrocorticographic signals as a proof-of-concept. Neural matched filters were created from monosyllabic consonant-vowel utterances: one keyword utterance, and 11 similar non-keyword utterances. These filters were used in an analog to the acoustic keyword spotting problem, applied for the first time to neural data. The filter templates were cross-correlated with the neural signal, capturing temporal dynamics of neural activation across cortical sites. Neural vocal activity detection (VAD) was used to identify utterance times and a discriminative classifier was used to determine if these utterances were the keyword or non-keyword speech. Model performance appeared to be highly related to electrode placement and spatial density. Vowel height (/a/ vs /i/) was poorly discriminated in recordings from sensorimotor cortex, but was highly discriminable using neural features from superior temporal gyrus during self-monitoring. The best performing neural keyword detection (5 keyword detections with two false-positives across 60 utterances) and neural VAD (100% sensitivity, ~1 false detection per 10 utterances) came from high-density (2 mm electrode diameter and 5 mm pitch) recordings from ventral sensorimotor cortex, suggesting the spatial fidelity and extent of high-density ECoG arrays may be sufficient for the purpose of speech brain-computer-interfaces.

Cortical Responses to Input From Distant Areas are Modulated by Local Spontaneous Alpha/Beta Oscillations.

Any given area in human cortex may receive input from multiple, functionally heterogeneous areas, potentially representing different processing threads. Alpha (8-13 Hz) and beta oscillations (13-20 Hz) have been hypothesized by other investigators to gate local cortical processing, but their influence on cortical responses to input from other cortical areas is unknown. To study this, we measured the effect of local oscillatory power and phase on cortical responses elicited by single-pulse electrical stimulation (SPES) at distant cortical sites, in awake human subjects implanted with intracranial electrodes for epilepsy surgery. In 4 out of 5 subjects, the amplitudes of corticocortical evoked potentials (CCEPs) elicited by distant SPES were reproducibly modulated by the power, but not the phase, of local oscillations in alpha and beta frequencies. Specifically, CCEP amplitudes were higher when average oscillatory power just before distant SPES (-110 to -10 ms) was high. This effect was observed in only a subset (0-33%) of sites with CCEPs and, like the CCEPs themselves, varied with stimulation at different distant sites. Our results suggest that although alpha and beta oscillations may gate local processing, they may also enhance the responsiveness of cortex to input from distant cortical sites.

Cortical subnetwork dynamics during human language tasks.

Language tasks require the coordinated activation of multiple subnetworks-groups of related cortical interactions involved in specific components of task processing. Although electrocorticography (ECoG) has sufficient temporal and spatial resolution to capture the dynamics of event-related interactions between cortical sites, it is difficult to decompose these complex spatiotemporal patterns into functionally discrete subnetworks without explicit knowledge of each subnetwork's timing. We hypothesized that subnetworks corresponding to distinct components of task-related processing could be identified as groups of interactions with co-varying strengths. In this study, five subjects implanted with ECoG grids over language areas performed word repetition and picture naming. We estimated the interaction strength between each pair of electrodes during each task using a time-varying dynamic Bayesian network (tvDBN) model constructed from the power of high gamma (70-110Hz) activity, a surrogate for population firing rates. We then reduced the dimensionality of this model using principal component analysis (PCA) to identify groups of interactions with co-varying strengths, which we term functional network components (FNCs). This data-driven technique estimates both the weight of each interaction's contribution to a particular subnetwork, and the temporal profile of each subnetwork's activation during the task. We found FNCs with temporal and anatomical features consistent with articulatory preparation in both tasks, and with auditory and visual processing in the word repetition and picture naming tasks, respectively. These FNCs were highly consistent between subjects with similar electrode placement, and were robust enough to be characterized in single trials. Furthermore, the interaction patterns uncovered by FNC analysis correlated well with recent literature suggesting important functional-anatomical distinctions between processing external and self-produced speech. Our results demonstrate that subnetwork decomposition of event-related cortical interactions is a powerful paradigm for interpreting the rich dynamics of large-scale, distributed cortical networks during human cognitive tasks.

Directed causality of the human electrocorticogram during dexterous movement.

While significant strides have been made in designing brain-machine interfaces for use in humans, efforts to decode truly dexterous movements in real time have been hindered by difficulty extracting detailed movement-related information from the most practical human neural interface, the electrocorticogram (ECoG). We explore a potentially rich, largely untapped source of movement-related information in the form of cortical connectivity computed with time-varying dynamic Bayesian networks (TV-DBN). We discover that measures of connectivity between ECoG electrodes derived from the local motor potential vary with dexterous movement in 65% of movement-related electrode pairs tested, and measures of connectivity derived from spectral features vary with dexterous movement in 76%. Due to the large number of features generated with connectivity methods, the TV-DBN a promising tool for dexterous decoding.