Crowdsourcing seizure detection: algorithm development and validation on human implanted device recordings

Combining general and personal models for epilepsy detection with hyperdimensional computing

Epilepsy is a highly prevalent chronic neurological disorder with great negative impact on patients' daily lives. Despite this there is still no adequate technological support to enable epilepsy detection and continuous outpatient monitoring in everyday life. Hyperdimensional (HD) computing is a promising method for epilepsy detection via wearable devices, characterized by a simpler learning process and lower memory requirements compared to other methods. In this work, we demonstrate additional avenues in which HD computing and the manner in which its models are built and stored can be used to better understand, compare and create more advanced machine learning models for epilepsy detection. These possibilities are not feasible with other state-of-the-art models, such as random forests or neural networks. We compare inter-subject model similarity of different classes (seizure and non-seizure), study the process of creating general models from personal ones, and finally posit a method of combining personal and general models to create hybrid models. This results in an improved epilepsy detection performance. We also tested knowledge transfer between models trained on two different datasets. The attained insights are highly interesting not only from an engineering perspective, to create better models for wearables, but also from a neurological perspective, to better understand individual epilepsy patterns.

EEG based automated seizure detection - A survey of medical professionals

Diagnosing and managing seizures presents substantial challenges for clinicians caring for patients with epilepsy. Although machine learning (ML) has been proposed for automated seizure detection using EEG data, there is little evidence of these technologies being broadly adopted in clinical practice. Moreover, there is a noticeable lack of surveys investigating this topic from the perspective of medical practitioners, which limits the understanding of the obstacles for the development of effective automated seizure detection. Besides the issue of generalisability and replicability seen in a small amount of studies, obstacles to the adoption of automated seizure detection remain largely unknown. To understand the obstacles preventing the application of seizure detection tools in clinical practice, we conducted a survey targeting medical professionals involved in the management of epilepsy. Our study aimed to gather insights on various factors such as the clinical utility, professional sentiment, benchmark requirements, and perceived barriers associated with the use of automated seizure detection tools. Our key findings are: I) The minimum acceptable sensitivity reported by most of our respondents (80%) seems achievable based on studies reported from most currently available ML-based EEG seizure detection algorithms, but replication studies often fail to meet this minimum. II) Respondents are receptive to the adoption of ML seizure detection tools and willing to spend time in training. III) The top three barriers for usage of such tools in clinical practice are related to availability, lack of training, and the blackbox nature of ML algorithms. Based on our findings, we developed a guide that can serve as a basis for developing ML-based seizure detection tools that meet the requirements of medical professionals, and foster the integration of these tools into clinical practice.

Artificial intelligence in epilepsy phenotyping

Artificial intelligence (AI) allows data analysis and integration at an unprecedented granularity and scale. Here we review the technological advances, challenges, and future perspectives of using AI for electro-clinical phenotyping of animal models and patients with epilepsy. In translational research, AI models accurately identify behavioral states in animal models of epilepsy, allowing identification of correlations between neural activity and interictal and ictal behavior. Clinical applications of AI-based automated and semi-automated analysis of audio and video recordings of people with epilepsy, allow significant data reduction and reliable detection and classification of major motor seizures. AI models can accurately identify electrographic biomarkers of epilepsy, such as spikes, high-frequency oscillations, and seizure patterns. Integrating AI analysis of electroencephalographic, clinical, and behavioral data will contribute to optimizing therapy for patients with epilepsy.

Automated sleep classification with chronic neural implants in freely behaving canines

Objective.Long-term intracranial electroencephalography (iEEG) in freely behaving animals provides valuable electrophysiological information and when correlated with animal behavior is useful for investigating brain function.Approach.Here we develop and validate an automated iEEG-based sleep-wake classifier for canines using expert sleep labels derived from simultaneous video, accelerometry, scalp electroencephalography (EEG) and iEEG monitoring. The video, scalp EEG, and accelerometry recordings were manually scored by a board-certified sleep expert into sleep-wake state categories: awake, rapid-eye-movement (REM) sleep, and three non-REM sleep categories (NREM1, 2, 3). The expert labels were used to train, validate, and test a fully automated iEEG sleep-wake classifier in freely behaving canines.Main results. The iEEG-based classifier achieved an overall classification accuracy of 0.878 ± 0.055 and a Cohen's Kappa score of 0.786 ± 0.090. Subsequently, we used the automated iEEG-based classifier to investigate sleep over multiple weeks in freely behaving canines. The results show that the dogs spend a significant amount of the day sleeping, but the characteristics of daytime nap sleep differ from night-time sleep in three key characteristics: during the day, there are fewer NREM sleep cycles (10.81 ± 2.34 cycles per day vs. 22.39 ± 3.88 cycles per night;p< 0.001), shorter NREM cycle durations (13.83 ± 8.50 min per day vs. 15.09 ± 8.55 min per night;p< 0.001), and dogs spend a greater proportion of sleep time in NREM sleep and less time in REM sleep compared to night-time sleep (NREM 0.88 ± 0.09, REM 0.12 ± 0.09 per day vs. NREM 0.80 ± 0.08, REM 0.20 ± 0.08 per night;p< 0.001).Significance.These results support the feasibility and accuracy of automated iEEG sleep-wake classifiers for canine behavior investigations.

Quantitative approaches to guide epilepsy surgery from intracranial EEG

Over the past 10 years, the drive to improve outcomes from epilepsy surgery has stimulated widespread interest in methods to quantitatively guide epilepsy surgery from intracranial EEG (iEEG). Many patients fail to achieve seizure freedom, in part due to the challenges in subjective iEEG interpretation. To address this clinical need, quantitative iEEG analytics have been developed using a variety of approaches, spanning studies of seizures, interictal periods, and their transitions, and encompass a range of techniques including electrographic signal analysis, dynamical systems modeling, machine learning and graph theory. Unfortunately, many methods fail to generalize to new data and are sensitive to differences in pathology and electrode placement. Here, we critically review selected literature on computational methods of identifying the epileptogenic zone from iEEG. We highlight shared methodological challenges common to many studies in this field and propose ways that they can be addressed. One fundamental common pitfall is a lack of open-source, high-quality data, which we specifically address by sharing a centralized high-quality, well-annotated, multicentre dataset consisting of >100 patients to support larger and more rigorous studies. Ultimately, we provide a road map to help these tools reach clinical trials and hope to improve the lives of future patients.

EEG datasets for seizure detection and prediction- A review

Electroencephalogram (EEG) datasets from epilepsy patients have been used to develop seizure detection and prediction algorithms using machine learning (ML) techniques with the aim of implementing the learned model in a device. However, the format and structure of publicly available datasets are different from each other, and there is a lack of guidelines on the use of these datasets. This impacts the generatability, generalizability, and reproducibility of the results and findings produced by the studies. In this narrative review, we compiled and compared the different characteristics of the publicly available EEG datasets that are commonly used to develop seizure detection and prediction algorithms. We investigated the advantages and limitations of the characteristics of the EEG datasets. Based on our study, we identified 17 characteristics that make the EEG datasets unique from each other. We also briefly looked into how certain characteristics of the publicly available datasets affect the performance and outcome of a study, as well as the influences it has on the choice of ML techniques and preprocessing steps required to develop seizure detection and prediction algorithms. In conclusion, this study provides a guideline on the choice of publicly available EEG datasets to both clinicians and scientists working to develop a reproducible, generalizable, and effective seizure detection and prediction algorithm.

Seizure forecasting using machine learning models trained by seizure diaries

Objectives.People with refractory epilepsy are overwhelmed by the uncertainty of their next seizures. Accurate prediction of future seizures could greatly improve the quality of life for these patients. New evidence suggests that seizure occurrences can have cyclical patterns for some patients. Even though these cyclicalities are not intuitive, they can be identified by machine learning (ML), to identify patients with predictable vs unpredictable seizure patterns.Approach.Self-reported seizure logs of 153 patients from the Human Epilepsy Project with more than three reported seizures (totaling 8337 seizures) were used to obtain inter-seizure interval time-series for training and evaluation of the forecasting models. Two classes of prediction methods were studied: (1) statistical approaches using Bayesian fusion of population-wise and individual-wise seizure patterns; and (2) ML-based algorithms including least squares, least absolute shrinkage and selection operator, support vector machine (SVM) regression, and long short-term memory regression. Leave-one-person-out cross-validation was used for training and evaluation, by training on seizure diaries of all except one subject and testing on the left-out subject.Main results.The leading forecasting models were the SVM regression and a statistical model that combined the median of population-wise seizure time-intervals with a test subject's prior seizure intervals. SVM was able to forecast 50%, 70%, 81%, 84%, and 87% of seizures of unseen subjects within 0, 1, 2, 3 to 4 d of mean absolute forecasting error, respectively. The subject-wise performances show that patients with more frequent seizures were generally better predicted.Significance.ML models can leverage non-random patterns within self-reported seizure diaries to forecast future seizures. While diary-based seizure forecasting alone is only one of many aspects of clinical care of patients with epilepsy, studying the level of predictability across seizures and patients paves the path towards a better understanding of predictable vs unpredictable seizures on individualized and population-wise bases.

Seizure count forecasting to aid diagnostic testing in epilepsy

OBJECTIVE

Epilepsy monitoring unit (EMU) admissions are critical for presurgical evaluation of drug-resistant epilepsy but may be nondiagnostic if an insufficient number of seizures are recorded. Seizure forecasting algorithms have shown promise for estimating the likelihood of seizures as a binary event in individual patients, but methods to predict how many seizures will occur remain elusive. Such methods could increase the diagnostic yield of EMU admissions and help patients mitigate seizure-related morbidity. Here, we evaluated the performance of a state-space method that uses prior seizure count data to predict future counts.

METHODS

A Bayesian negative-binomial dynamic linear model (DLM) was developed to forecast daily electrographic seizure counts in 19 patients implanted with a responsive neurostimulation (RNS) device. Holdout validation was used to evaluate performance in predicting the number of electrographic seizures for forecast horizons ranging 1-7 days ahead.

RESULTS

One-day-ahead prediction of the number of electrographic seizures using a negative-binomial DLM resulted in improvement over chance in 73.1% of time segments compared to a random chance forecaster and remained >50% for forecast horizons of up to 7 days. Superior performance (mean error = .99) was obtained in predicting the number of electrographic seizures in the next day compared to three traditional methods for count forecasting (integer-valued generalized autoregressive conditional heteroskedasticity model or INGARCH, 1.10; Croston, 1.06; generalized linear autoregressive moving average model or GLARMA, 2.00). Number of electrographic seizures in the preceding day and laterality of electrographic pattern detections had highest predictive value, with greater number of electrographic seizures and RNS magnet swipes in the preceding day associated with a higher number of electrographic seizures the next day.

SIGNIFICANCE

This study demonstrates that DLMs can predict the number of electrographic seizures a patient will experience days in advance with above chance accuracy. This study represents an important step toward the translation of seizure forecasting methods into the optimization of EMU admissions.

Novel subscalp and intracranial devices to wirelessly record and analyze continuous EEG in unsedated, behaving dogs in their natural environments: A new paradigm in canine epilepsy research

Epilepsy is characterized by unprovoked, recurrent seizures and is a common neurologic disorder in dogs and humans. Roughly 1/3 of canines and humans with epilepsy prove to be drug-resistant and continue to have sporadic seizures despite taking daily anti-seizure medications. The optimization of pharmacologic therapy is often limited by inaccurate seizure diaries and medication side effects. Electroencephalography (EEG) has long been a cornerstone of diagnosis and classification in human epilepsy, but because of several technical challenges has played a smaller clinical role in canine epilepsy. The interictal (between seizures) and ictal (seizure) EEG recorded from the epileptic mammalian brain shows characteristic electrophysiologic biomarkers that are very useful for clinical management. A fundamental engineering gap for both humans and canines with epilepsy has been the challenge of obtaining continuous long-term EEG in the patients' natural environment. We are now on the cusp of a revolution where continuous long-term EEG from behaving canines and humans will be available to guide clinicians in the diagnosis and optimal treatment of their patients. Here we review some of the devices that have recently emerged for obtaining long-term EEG in ambulatory subjects living in their natural environments.

Exploring Novel Innovation Strategies to Close a Technology Gap in Neurosurgery: The HORAO Crowdsourcing Campaign (Preprint)

BACKGROUND

Scientific research is typically performed by expert individuals or groups who investigate potential solutions in a sequential manner. Given the current worldwide exponential increase in technical innovations, potential solutions for any new problem might already exist, even though they were developed to solve a different problem. Therefore, in crowdsourcing ideation, a research question is explained to a much larger group of individuals beyond the specialist community to obtain a multitude of diverse, outside-the-box solutions. These are then assessed in parallel by a group of experts for their capacity to solve the new problem. The 2 key problems in brain tumor surgery are the difficulty of discerning the exact border between a tumor and the surrounding brain, and the difficulty of identifying the function of a specific area of the brain. Both problems could be solved by a method that visualizes the highly organized fiber tracts within the brain; the absence of fibers would reveal the tumor, whereas the spatial orientation of the tracts would reveal the area's function. To raise awareness about our challenge of developing a means of intraoperative, real-time, noninvasive identification of fiber tracts and tumor borders to improve neurosurgical oncology, we turned to the crowd with a crowdsourcing ideation challenge.

OBJECTIVE

Our objective was to evaluate the feasibility of a crowdsourcing ideation campaign for finding novel solutions to challenges in neuroscience. The purpose of this paper is to introduce our chosen crowdsourcing method and discuss it in the context of the current literature.

METHODS

We ran a prize-based crowdsourcing ideation competition called HORAO on the commercial platform HeroX. Prize money previously collected through a crowdfunding campaign was offered as an incentive. Using a multistage approach, an expert jury first selected promising technical solutions based on broad, predefined criteria, coached the respective solvers in the second stage, and finally selected the winners in a conference setting. We performed a postchallenge web-based survey among the solvers crowd to find out about their backgrounds and demographics.

RESULTS

Our web-based campaign reached more than 20,000 people (views). We received 45 proposals from 32 individuals and 7 teams, working in 26 countries on 4 continents. The postchallenge survey revealed that most of the submissions came from single solvers or teams working in engineering or the natural sciences, with additional submissions from other nonmedical fields. We engaged in further exchanges with 3 out of the 5 finalists and finally initiated a successful scientific collaboration with the winner of the challenge.

CONCLUSIONS

This open innovation competition is the first of its kind in medical technology research. A prize-based crowdsourcing ideation campaign is a promising strategy for raising awareness about a specific problem, finding innovative solutions, and establishing new scientific collaborations beyond strictly disciplinary domains.

Distributed brain co-processor for tracking spikes, seizures and behavior during electrical brain stimulation

Early implantable epilepsy therapy devices provided open-loop electrical stimulation without brain sensing, computing, or an interface for synchronized behavioral inputs from patients. Recent epilepsy stimulation devices provide brain sensing but have not yet developed analytics for accurately tracking and quantifying behavior and seizures. Here we describe a distributed brain co-processor providing an intuitive bi-directional interface between patient, implanted neural stimulation and sensing device, and local and distributed computing resources. Automated analysis of continuous streaming electrophysiology is synchronized with patient reports using a hand-held device and integrated with distributed cloud computing resources for quantifying seizures, interictal epileptiform spikes, and patient symptoms during therapeutic electrical brain stimulation. The classification algorithms for interictal epileptiform spikes and seizures were developed and parameterized using long-term ambulatory data from 9 humans and 8 canines with epilepsy, and then implemented prospectively in out-of-sample testing in 2 pet canines and 4 humans with drug resistant epilepsy living in their natural environments. Accurate seizure diaries are needed as the primary clinical outcome measure of epilepsy therapy and to guide brain stimulation optimization. The brain co-processor system described here enables tracking interictal epileptiform spikes, seizures, and correlation with patient behavioral reports. In the future correlation of spikes and seizures with behavior will allow more detailed investigation of the clinical impact of spikes and seizures on patients.

Dynamical Network Models From EEG and MEG for Epilepsy Surgery—A Quantitative Approach

There is an urgent need for more informative quantitative techniques that non-invasively and objectively assess strategies for epilepsy surgery. Invasive intracranial electroencephalography (iEEG) remains the clinical gold standard to investigate the nature of the epileptogenic zone (EZ) before surgical resection. However, there are major limitations of iEEG, such as the limited spatial sampling and the degree of subjectivity inherent in the analysis and clinical interpretation of iEEG data. Recent advances in network analysis and dynamical network modeling provide a novel aspect toward a more objective assessment of the EZ. The advantage of such approaches is that they are data-driven and require less or no human input. Multiple studies have demonstrated success using these approaches when applied to iEEG data in characterizing the EZ and predicting surgical outcomes. However, the limitations of iEEG recordings equally apply to these studies—limited spatial sampling and the implicit assumption that iEEG electrodes, whether strip, grid, depth or stereo EEG (sEEG) arrays, are placed in the correct location. Therefore, it is of interest to clinicians and scientists to see whether the same analysis and modeling techniques can be applied to whole-brain, non-invasive neuroimaging data (from MRI-based techniques) and neurophysiological data (from MEG and scalp EEG recordings), thus removing the limitation of spatial sampling, while safely and objectively characterizing the EZ. This review aims to summarize current state of the art non-invasive methods that inform epilepsy surgery using network analysis and dynamical network models. We also present perspectives on future directions and clinical applications of these promising approaches.

A Comparison of Energy-Efficient Seizure Detectors for Implantable Neurostimulation Devices

Introduction

About 30% of epilepsy patients are resistant to treatment with antiepileptic drugs, and only a minority of these are surgical candidates. A recent therapeutic approach is the application of electrical stimulation in the early phases of a seizure to interrupt its spread across the brain. To accomplish this, energy-efficient seizure detectors are required that are able to detect a seizure in its early stages.

Methods

Three patient-specific, energy-efficient seizure detectors are proposed in this study: (i) random forest (RF); (ii) long short-term memory (LSTM) recurrent neural network (RNN); and (iii) convolutional neural network (CNN). Performance evaluation was based on EEG data (n = 40 patients) derived from a selected set of surface EEG electrodes, which mimic the electrode layout of an implantable neurostimulation system. As for the RF input, 16 features in the time- and frequency-domains were selected. Raw EEG data were used for both CNN and RNN. Energy consumption was estimated by a platform-independent model based on the number of arithmetic operations (AOs) and memory accesses (MAs). To validate the estimated energy consumption, the RNN classifier was implemented on an ultra-low-power microcontroller.

Results

The RNN seizure detector achieved a slightly better level of performance, with a median area under the precision-recall curve score of 0.49, compared to 0.47 for CNN and 0.46 for RF. In terms of energy consumption, RF was the most efficient algorithm, with a total of 67k AOs and 67k MAs per classification. This was followed by CNN (488k AOs and 963k MAs) and RNN (772k AOs and 978k MAs), whereby MAs contributed more to total energy consumption. Measurements derived from the hardware implementation of the RNN algorithm demonstrated a significant correlation between estimations and actual measurements.

Discussion

All three proposed seizure detection algorithms were shown to be suitable for application in implantable devices. The applied methodology for a platform-independent energy estimation was proven to be accurate by way of hardware implementation of the RNN algorithm. These findings show that seizure detection can be achieved using just a few channels with limited spatial distribution. The methodology proposed in this study can therefore be applied when designing new models for responsive neurostimulation.

Thalamic deep brain stimulation modulates cycles of seizure risk in epilepsy

Chronic brain recordings suggest that seizure risk is not uniform, but rather varies systematically relative to daily (circadian) and multiday (multidien) cycles. Here, one human and seven dogs with naturally occurring epilepsy had continuous intracranial EEG (median 298 days) using novel implantable sensing and stimulation devices. Two pet dogs and the human subject received concurrent thalamic deep brain stimulation (DBS) over multiple months. All subjects had circadian and multiday cycles in the rate of interictal epileptiform spikes (IES). There was seizure phase locking to circadian and multiday IES cycles in five and seven out of eight subjects, respectively. Thalamic DBS modified circadian (all 3 subjects) and multiday (analysis limited to the human participant) IES cycles. DBS modified seizure clustering and circadian phase locking in the human subject. Multiscale cycles in brain excitability and seizure risk are features of human and canine epilepsy and are modifiable by thalamic DBS.

Simulated diagnostic performance of low-field MRI: Harnessing open-access datasets to evaluate novel devices

The purpose of this study is to demonstrate a method for virtually evaluating novel imaging devices using machine learning and open-access datasets, here applied to a new, low-field strength portable 64mT MRI device. Paired 3 T and 64mT brain images were used to develop and validate a transformation converting standard clinical images to low-field quality images. Separately, 3 T images were aggregated from open-source databases spanning four neuropathologies: low-grade glioma (LGG, N = 76), high-grade glioma (HGG, N = 259), stroke (N = 28), and multiple sclerosis (MS, N = 20). The transformation method was then applied to the open-source data to generate simulated low-field images for each pathology. Convolutional neural networks (DenseNet-121) were trained to detect pathology in axial slices from either 3 T or simulated 64 mT images, and their relative performance was compared to characterize the potential diagnostic capabilities of low-field imaging. Algorithm performance was measured using area under the receiver operating characteristic curve. Across all cohorts, pathology detection was similar between 3 T and simulated 64mT images (LGG: 0.97 vs. 0.98; HGG: 0.96 vs. 0.95; stroke: 0.94 vs. 0.94; MS: 0.90 vs 0.87). Pathology detection was further characterized as a function of lesion size, intensity, and contrast. Simulated images showed decreasing sensitivity for lesions smaller than 4 cm². While simulations cannot replace prospective trials during the evaluation of medical devices, they can provide guidance and justification for prospective studies. Simulated data derived from open-source imaging databases may facilitate testing and validation of new imaging devices.

An Overview of EEG-based Machine Learning Methods in Seizure Prediction and Opportunities for Neurologists in this Field

The unpredictability of epileptic seizures is one of the most problematic aspects of the field of epilepsy. Methods or devices capable of detecting seizures minutes before they occur may help prevent injury or even death and significantly improve the quality of life. Machine learning (ML) is an emerging technology that can markedly enhance algorithm performance by interpreting data. ML has gained increasing attention from medical researchers in recent years. Its epilepsy applications range from the localization of the epileptic region, predicting the medical or surgical outcome of epilepsy, and automated electroencephalography (EEG) analysis to seizure prediction. While ML has good prospects with regard to detecting epileptic seizures via EEG signals, many clinicians are still unfamiliar with this field. This work briefly summarizes the history and recent significant progress made in this field and clarifies the essential components of the automatic seizure detection system using ML methodologies for clinicians. This review also proposes how neurologists can actively contribute to ensure improvements in seizure prediction using EEG-based ML.

Wearable Reduced-Channel EEG System for Remote Seizure Monitoring

Epitel has developed Epilog, a miniature, wireless, wearable electroencephalography (EEG) sensor. Four Epilog sensors are combined as part of Epitel's Remote EEG Monitoring platform (REMI) to create 10 channels of EEG for remote patient monitoring. REMI is designed to provide comprehensive spatial EEG recordings that can be administered by non-specialized medical personnel in any medical center. The purpose of this study was to determine how accurate epileptologists are at remotely reviewing Epilog sensor EEG in the 10-channel “REMI montage,” with and without seizure detection support software. Three board certified epileptologists reviewed the REMI montage from 20 subjects who wore four Epilog sensors for up to 5 days alongside traditional video-EEG in the EMU, 10 of whom experienced a total of 24 focal-onset electrographic seizures and 10 of whom experienced no seizures or epileptiform activity. Epileptologists randomly reviewed the same datasets with and without clinical decision support annotations from an automated seizure detection algorithm tuned to be highly sensitive. Blinded consensus review of unannotated Epilog EEG in the REMI montage detected people who were experiencing electrographic seizure activity with 90% sensitivity and 90% specificity. Consensus detection of individual focal onset seizures resulted in a mean sensitivity of 61%, precision of 80%, and false detection rate (FDR) of 0.002 false positives per hour (FP/h) of data. With algorithm seizure detection annotations, the consensus review mean sensitivity improved to 68% with a slight increase in FDR (0.005 FP/h). As seizure detection software, the automated algorithm detected people who were experiencing electrographic seizure activity with 100% sensitivity and 70% specificity, and detected individual focal onset seizures with a mean sensitivity of 90% and mean false alarm rate of 0.087 FP/h. This is the first study showing epileptologists' ability to blindly review EEG from four Epilog sensors in the REMI montage, and the results demonstrate the clinical potential to accurately identify patients experiencing electrographic seizures. Additionally, the automated algorithm shows promise as clinical decision support software to detect discrete electrographic seizures in individual records as accurately as FDA-cleared predicates.

Phase-Amplitude Coupling Features Accurately Classify Multiple Sub-States Within a Seizure Episode

Epilepsy is frequently characterized by convulsive seizures, which are often followed by a postictal EEG suppression state (PGES). The ability to automatically detect and monitor seizure progression and postictal state can allow for early warning of seizure onset, timely intervention in seizures themselves, as well as identification of major complications in epilepsy such as status epilepticus and sudden unexpected death in epilepsy (SUDEP). To test whether it is possible to reliably differentiate these ictal and postictal states, we investigated 52 seizure records (both intracranial and scalp EEG) from 19 patients. Phase-amplitude cross-frequency coupling was calculated for each recording and used as an input to a convolutional neural network model, achieving the mean accuracy of 0.890.09 across all classes, with the worst class accuracy of 0.73 for one of the later ictal sub-states. When the trained model was applied to SUDEP patient data, it classified seizure recordings as primarily interictal and PGES-like state (70% and 26%, respectively), highlighting the fact that in SUDEP patients seizures primarily exist in postictal states and don't show the ictal sub-state evolution. These results suggest that using frequency coupling markers with a machine learning algorithm can reliably identify ictal and postictal sub-states, which can open up opportunities for novel monitoring and management approaches in epilepsy.

Electrical Brain Stimulation for Epilepsy and Emerging Applications

SUMMARY

Electrical brain stimulation is an established therapy for movement disorders, epilepsy, obsessive compulsive disorder, and a potential therapy for many other neurologic and psychiatric disorders. Despite significant progress and FDA approvals, there remain significant clinical gaps that can be addressed with next generation systems. Integrating wearable sensors and implantable brain devices with off-the-body computing resources (smart phones and cloud resources) opens a new vista for dense behavioral and physiological signal tracking coupled with adaptive stimulation therapy that should have applications for a range of brain and mind disorders. Here, we briefly review some history and current electrical brain stimulation applications for epilepsy, deep brain stimulation and responsive neurostimulation, and emerging applications for next generation devices and systems.

Anterior nucleus of the thalamus seizure detection in ambulatory humans

There is a paucity of data to guide anterior nucleus of the thalamus (ANT) deep brain stimulation (DBS) with brain sensing. The clinical Medtronic Percept DBS device provides constrained brain sensing power within a frequency band (power-in-band [PIB]), recorded in 10-min averaged increments. Here, four patients with temporal lobe epilepsy were implanted with an investigational device providing full bandwidth chronic intracranial electroencephalogram (cEEG) from bilateral ANT and hippocampus (Hc). ANT PIB-based seizure detection was assessed. Detection parameters were cEEG PIB center frequency, bandwidth, and epoch duration. Performance was evaluated against epileptologist-confirmed Hc seizures, and assessed by area under the precision-recall curve (PR-AUC). Data included 99 days of cEEG, and 20, 278, 3, and 18 Hc seizures for Subjects 1-4. The best detector had 7-Hz center frequency, 5-Hz band width, and 10-s epoch duration (group PR-AUC = .90), with 75% sensitivity and .38 false alarms per day for Subject 1, and 100% and .0 for Subjects 3 and 4. Hc seizures in Subject 2 did not propagate to ANT. The relative change of ANT PIB was maximal ipsilateral to seizure onset for all detected seizures. Chronic ANT and Hc recordings provide direct guidance for ANT DBS with brain sensing.

Data-driven electrophysiological feature based on deep learning to detect epileptic seizures

Objective. To identify a new electrophysiological feature characterising the epileptic seizures, which is commonly observed in different types of epilepsy.Methods. We recorded the intracranial electroencephalogram (iEEG) of 21 patients (12 women and 9 men) with multiple types of refractory epilepsy. The raw iEEG signals of the early phase of epileptic seizures and interictal states were classified by a convolutional neural network (Epi-Net). For comparison, the same signals were classified by a support vector machine (SVM) using the spectral power and phase-amplitude coupling. The features learned by Epi-Net were derived by a modified integrated gradients method. We considered the product of powers multiplied by the relative contribution of each frequency amplitude as a data-driven epileptogenicity index (d-EI). We compared the d-EI and other conventional features in terms of accuracy to detect the epileptic seizures. Finally, we compared the d-EI among the electrodes to evaluate its relationship with the resected area and the Engel classification.Results. Epi-Net successfully identified the epileptic seizures, with an area under the receiver operating characteristic curve of 0.944 ± 0.067, which was significantly larger than that of the SVM (0.808 ± 0.253,n =21;p =0.025). The learned iEEG signals were characterised by increased powers of 17-92 Hz and >180 Hz in addition to decreased powers of other frequencies. The proposed d-EI detected them with better accuracy than the other iEEG features. Moreover, the surgical resection of areas with a larger increase in d-EI was observed for all nine patients with Engel class ⩽1, but not for the 4 of 12 patients with Engel class >1, demonstrating the significant association with seizure outcomes.Significance.We derived an iEEG feature from the trained Epi-Net, which identified the epileptic seizures with improved accuracy and might contribute to identification of the epileptogenic zone.

Effects of anterior thalamic nuclei stimulation on hippocampal activity: Chronic recording in a patient with drug-resistant focal epilepsy

•

Devices for RNS and thalamic DBS were implanted in a single person with epilepsy.

•

RNS electrocorticography enabled characterization of acute and chronic DBS effects.

•

DBS caused acute, phasic, frequency-dependent responses in hippocampus and cortex.

•

DBS modulated functional connectivity and suppressed epileptiform activity over time.

•

Chronic electrocorticography elucidates progressive effects of thalamic stimulation.

Implanted neurostimulation devices are gaining traction as palliative treatment options for certain forms of drug-resistant epilepsy, but clinical utility of these devices is hindered by incomplete mechanistic understanding of their therapeutic effects. Approved devices for anterior thalamic nuclei deep brain stimulation (ANT DBS) are thought to work at a network level, but limited sensing capability precludes characterization of neurophysiological effects outside the thalamus. Here, we describe a patient with drug-resistant temporal lobe epilepsy who was implanted with a responsive neurostimulation device (RNS System), involving hippocampal and ipsilateral temporal neocortical leads, and subsequently received ANT DBS. Over 1.5 years, RNS System electrocorticography enabled multiscale characterization of neurophysiological effects of thalamic stimulation. In brain regions sampled by the RNS System, ANT DBS produced acute, phasic, frequency-dependent responses, including suppression of hippocampal low frequency local field potentials. ANT DBS modulated functional connectivity between hippocampus and neocortex. Finally, ANT DBS progressively suppressed hippocampal epileptiform activity in relation to the extent of hippocampal theta suppression, which informs stimulation parameter selection for ANT DBS. Taken together, this unique clinical scenario, involving hippocampal recordings of unprecedented chronicity alongside ANT DBS, sheds light on the therapeutic mechanism of thalamic stimulation and highlights capabilities needed in next-generation devices.

Wearable devices for seizure detection: Practical experiences and recommendations from the Wearables for Epilepsy And Research (WEAR) International Study Group

The Wearables for Epilepsy And Research (WEAR) International Study Group identified a set of methodology standards to guide research on wearable devices for seizure detection. We formed an international consortium of experts from clinical research, engineering, computer science, and data analytics at the beginning of 2020. The study protocols and practical experience acquired during the development of wearable research studies were discussed and analyzed during bi-weekly virtual meetings to highlight commonalities, strengths, and weaknesses, and to formulate recommendations. Seven major essential components of the experimental design were identified, and recommendations were formulated about: (1) description of study aims, (2) policies and agreements, (3) study population, (4) data collection and technical infrastructure, (5) devices, (6) reporting results, and (7) data sharing. Introducing a framework of methodology standards promotes optimal, accurate, and consistent data collection. It also guarantees that studies are generalizable and comparable, and that results can be replicated, validated, and shared.

Computer-assisted EEG diagnostic review for idiopathic generalized epilepsy

Epilepsy diagnosis can be costly, time-consuming, and not uncommonly inaccurate. The reference standard diagnostic monitoring is continuous video-electroencephalography (EEG) monitoring, ideally capturing all events or concordant interictal discharges. Automating EEG data review would save time and resources, thus enabling more people to receive reference standard monitoring and also potentially heralding a more quantitative approach to therapeutic outcomes. There is substantial research into the automated detection of seizures and epileptic activity from EEG. However, automated detection software is not widely used in the clinic, and despite numerous published algorithms, few methods have regulatory approval for detecting epileptic activity from EEG. This study reports on a deep learning algorithm for computer-assisted EEG review. Deep convolutional neural networks were trained to detect epileptic discharges using a preexisting dataset of over 6000 labelled events in a cohort of 103 patients with idiopathic generalized epilepsy (IGE). Patients underwent 24-hour ambulatory outpatient EEG, and all data were curated and confirmed independently by two epilepsy specialists (Seneviratne et al., 2016). The resulting automated detection algorithm was then used to review diagnostic scalp EEG for seven patients (four with IGE and three with events mimicking seizures) to validate performance in a clinical setting. The automated detection algorithm showed state-of-the-art performance for detecting epileptic activity from clinical EEG, with mean sensitivity of >95% and corresponding mean false positive rate of 1 detection per minute. Importantly, diagnostic case studies showed that the automated detection algorithm reduced human review time by 80%-99%, without compromising event detection or diagnostic accuracy. The presented results demonstrate that computer-assisted review can increase the speed and accuracy of EEG assessment and has the potential to greatly improve therapeutic outcomes. This article is part of the Special Issue "NEWroscience 2018".

Epilepsy Personal Assistant Device-A Mobile Platform for Brain State, Dense Behavioral and Physiology Tracking and Controlling Adaptive Stimulation

Epilepsy is one of the most common neurological disorders, and it affects almost 1% of the population worldwide. Many people living with epilepsy continue to have seizures despite anti-epileptic medication therapy, surgical treatments, and neuromodulation therapy. The unpredictability of seizures is one of the most disabling aspects of epilepsy. Furthermore, epilepsy is associated with sleep, cognitive, and psychiatric comorbidities, which significantly impact the quality of life. Seizure predictions could potentially be used to adjust neuromodulation therapy to prevent the onset of a seizure and empower patients to avoid sensitive activities during high-risk periods. Long-term objective data is needed to provide a clearer view of brain electrical activity and an objective measure of the efficacy of therapeutic measures for optimal epilepsy care. While neuromodulation devices offer the potential for acquiring long-term data, available devices provide very little information regarding brain activity and therapy effectiveness. Also, seizure diaries kept by patients or caregivers are subjective and have been shown to be unreliable, in particular for patients with memory-impairing seizures. This paper describes the design, architecture, and development of the Mayo Epilepsy Personal Assistant Device (EPAD). The EPAD has bi-directional connectivity to the implanted investigational Medtronic Summit RC+STM device to implement intracranial EEG and physiological monitoring, processing, and control of the overall system and wearable devices streaming physiological time-series signals. In order to mitigate risk and comply with regulatory requirements, we developed a Quality Management System (QMS) to define the development process of the EPAD system, including Risk Analysis, Verification, Validation, and protocol mitigations. Extensive verification and validation testing were performed on thirteen canines and benchtop systems. The system is now under a first-in-human trial as part of the US FDA Investigational Device Exemption given in 2018 to study modulated responsive and predictive stimulation using the Mayo EPAD system and investigational Medtronic Summit RC+STM in ten patients with non-resectable dominant or bilateral mesial temporal lobe epilepsy. The EPAD system coupled with an implanted device capable of EEG telemetry represents a next-generation solution to optimizing neuromodulation therapy.

A Full-Stack Application for Detecting Seizures and Reducing Data During Continuous Electroencephalogram Monitoring

Supplemental Digital Content is available in the text.

BACKGROUND:

Continuous electroencephalogram monitoring is associated with lower mortality in critically ill patients; however, it is underused due to the resource-intensive nature of manually interpreting prolonged streams of continuous electroencephalogram data. Here, we present a novel real-time, machine learning–based alerting and monitoring system for epilepsy and seizures that dramatically reduces the amount of manual electroencephalogram review.

METHODS:

We developed a custom data reduction algorithm using a random forest and deployed it within an online cloud-based platform, which streams data and communicates interactively with caregivers via a web interface to display algorithm results. We developed real-time, machine learning–based alerting and monitoring system for epilepsy and seizures on continuous electroencephalogram recordings from 77 patients undergoing routine scalp ICU electroencephalogram monitoring and tested it on an additional 20 patients.

RESULTS

We achieved a mean seizure sensitivity of 84% in cross-validation and 85% in testing, as well as a mean specificity of 83% in cross-validation and 86% in testing, corresponding to a high level of data reduction. This study validates a platform for machine learning–assisted continuous electroencephalogram analysis and represents a meaningful step toward improving utility and decreasing cost of continuous electroencephalogram monitoring. We also make our high-quality annotated dataset of 97 ICU continuous electroencephalogram recordings public for others to validate and improve upon our methods.

Seizure Diaries and Forecasting With Wearables: Epilepsy Monitoring Outside the Clinic

It is a major challenge in clinical epilepsy to diagnose and treat a disease characterized by infrequent seizures based on patient or caregiver reports and limited duration clinical testing. The poor reliability of self-reported seizure diaries for many people with epilepsy is well-established, but these records remain necessary in clinical care and therapeutic studies. A number of wearable devices have emerged, which may be capable of detecting seizures, recording seizure data, and alerting caregivers. Developments in non-invasive wearable sensors to measure accelerometry, photoplethysmography (PPG), electrodermal activity (EDA), electromyography (EMG), and other signals outside of the traditional clinical environment may be able to identify seizure-related changes. Non-invasive scalp electroencephalography (EEG) and minimally invasive subscalp EEG may allow direct measurement of seizure activity. However, significant network and computational infrastructure is needed for continuous, secure transmission of data. The large volume of data acquired by these devices necessitates computer-assisted review and detection to reduce the burden on human reviewers. Furthermore, user acceptability of such devices must be a paramount consideration to ensure adherence with long-term device use. Such devices can identify tonic–clonic seizures, but identification of other seizure semiologies with non-EEG wearables is an ongoing challenge. Identification of electrographic seizures with subscalp EEG systems has recently been demonstrated over long (>6 month) durations, and this shows promise for accurate, objective seizure records. While the ability to detect and forecast seizures from ambulatory intracranial EEG is established, invasive devices may not be acceptable for many individuals with epilepsy. Recent studies show promising results for probabilistic forecasts of seizure risk from long-term wearable devices and electronic diaries of self-reported seizures. There may also be predictive value in individuals' symptoms, mood, and cognitive performance. However, seizure forecasting requires perpetual use of a device for monitoring, increasing the importance of the system's acceptability to users. Furthermore, long-term studies with concurrent EEG confirmation are lacking currently. This review describes the current evidence and challenges in the use of minimally and non-invasive devices for long-term epilepsy monitoring, the essential components in remote monitoring systems, and explores the feasibility to detect and forecast impending seizures via long-term use of these systems.

Epileptic Seizure Detection on an Ultra-Low-Power Embedded RISC-V Processor Using a Convolutional Neural Network

The treatment of refractory epilepsy via closed-loop implantable devices that act on seizures either by drug release or electrostimulation is a highly attractive option. For such implantable medical devices, efficient and low energy consumption, small size, and efficient processing architectures are essential. To meet these requirements, epileptic seizure detection by analysis and classification of brain signals with a convolutional neural network (CNN) is an attractive approach. This work presents a CNN for epileptic seizure detection capable of running on an ultra-low-power microprocessor. The CNN is implemented and optimized in MATLAB. In addition, the CNN is also implemented on a GAP8 microprocessor with RISC-V architecture. The training, optimization, and evaluation of the proposed CNN are based on the CHB-MIT dataset. The CNN reaches a median sensitivity of 90% and a very high specificity over 99% corresponding to a median false positive rate of 6.8 s per hour. After implementation of the CNN on the microcontroller, a sensitivity of 85% is reached. The classification of 1 s of EEG data takes t=35 ms and consumes an average power of P≈140 μW. The proposed detector outperforms related approaches in terms of power consumption by a factor of 6. The universal applicability of the proposed CNN based detector is verified with recording of epileptic rats. This results enable the design of future medical devices for epilepsy treatment.

Prediction of Seizure Recurrence. A Note of Caution

Great strides have been made recently in documenting that machine-learning programs can predict seizure occurrence in people who have epilepsy. Along with this progress have come claims that appear to us to be a bit premature. We anticipate that many people will benefit from seizure prediction. We also doubt that all will benefit. Although machine learning is a useful tool for aiding discovery, we believe that the greatest progress will come from deeper understanding of seizures, epilepsy, and the EEG features that enable seizure prediction. In this essay, we lay out reasons for optimism and skepticism.

Validation of an EEG seizure detection paradigm optimized for clinical use in a chronically implanted subcutaneous device

BACKGROUND

Many electroencephalography (EEG) based seizure detection paradigms have been developed and validated over the last two decades. The majority of clinical approaches use scalp or intracranial EEG electrodes. Scalp EEG is limited by patient discomfort and short duration of useful EEG signals. Intracranial EEG involves an invasive surgical procedure associated with significant risk making it unsuitable for widespread use as a practical clinical biometric. A less invasive EEG monitoring approach, that is between invasive intracranial procedures and noninvasive methods, would fill the need of a safe, accurate, chronic (ultra-long term) and objective seizure detection method. We present validation of a continuous EEG seizure detection paradigm using human single-channel EEG recordings from subcutaneously placed electrodes that could be used to fulfill this need.

METHODS

Ten-minute long sleep, awake and ictal EEG epochs obtained from 21 human subjects with subscalp electrodes and validated against simultaneous iEEG recordings were analyzed by three experienced clinical neurophysiologists. The 201subscalp EEG time series epochs where classified as diagnostic for awake, asleep, or seizure by the clinicians who were blinded to all other information. Seventy of the epochs were classified in this way as representing seizure activity. A subject specific seizure detection algorithm was trained and then evaluated offline for each patient in the data set using the expert consensus classification as the gold standard.

RESULTS

The average seizure detection performance of the algorithm across 21 subjects exceeded 90 % accuracy: 97 % sensitivity, 91 % specificity, and 93 % accuracy. For 19 of 21 patient datasets the algorithm achieved 100 % sensitivity. For 15 of 21 patients, the algorithm achieved 100 % specificity. For 13 of 21 patients the algorithm achieved 100 % accuracy.

COMPARISON

No comparable published methods are available for subgaleal EEG seizure detection.

CONCLUSIONS

These findings suggest that a simple seizure detection algorithm using subcutaneous EEG signals could provide sufficient accuracy and clinical utility for use in a low power, long-term subcutaneous brain monitoring device. Such a device would fill a need for a large number of people with epilepsy who currently have no means for accurately quantifying their seizures thereby providing important information to healthcare providers not currently available.

Non-invasive wearable seizure detection using long-short-term memory networks with transfer learning

Objective. The detection of seizures using wearable devices would improve epilepsy management, but reliable detection of seizures in an ambulatory environment remains challenging, and current studies lack concurrent validation of seizures using electroencephalography (EEG) data.Approach. An adaptively trained long-short-term memory deep neural network was developed and trained using a modest number of seizure data sets from wrist-worn devices. Transfer learning was used to adapt a classifier that was initially trained on intracranial electroencephalography (iEEG) signals to facilitate classification of non-EEG physiological datasets comprising accelerometry, blood volume pulse, skin electrodermal activity, heart rate, and temperature signals. The algorithm's performance was assessed with and without pre-training on iEEG signals and transfer learning. To assess the performance of the seizure detection classifier using long-term ambulatory data, wearable devices were used for multiple months with an implanted neurostimulator capable of recording iEEG signals, which provided independent electrographic seizure detections that were reviewed by a board-certified epileptologist.Main results. For 19 motor seizures from 10 in-hospital patients, the algorithm yielded a mean area under curve (AUC), a sensitivity, and an false alarm rate per day (FAR/day) of 0.98, 0.93, and 2.3, respectively. Additionally, for eight seizures with probable motor semiology from two ambulatory patients, the classifier achieved a mean AUC of 0.97 and an FAR of 2.45 events/day at a sensitivity of 0.9. For all seizure types in the ambulatory setting, the classifier had a mean AUC of 0.82 with a sensitivity of 0.47 and an FAR of 7.2 events/day.Significance. The performance of the algorithm was evaluated using motor and non-motor seizures during in-hospital and ambulatory use. The classifier was able to detect multiple types of motor and non-motor seizures, but performed significantly better on motor seizures.

An Ensemble of Hyperdimensional Classifiers: Hardware-Friendly Short-Latency Seizure Detection With Automatic iEEG Electrode Selection

We propose a new algorithm for detecting epileptic seizures. Our algorithm first extracts three features, namely mean amplitude, line length, and local binary patterns that are fed to an ensemble of classifiers using hyperdimensional (HD) computing. These features are embedded into prototype vectors representing ictal (during seizures) and interictal (between seizures) brain states are constructed. These vectors can be computed at different spatial scales ranging from a single electrode up to many electrodes. This flexibility allows our algorithm to identify the electrodes that discriminate best between ictal and interictal brain states. We assess our algorithm on the SWEC-ETHZ iEEG dataset that includes 99 short-time iEEG seizures recorded with 36 to 100 electrodes from 16 drug-resistant epilepsy patients. Using k-fold cross-validation and all electrodes, our algorithm surpasses state-of-the-art algorithms yielding significantly shorter latency (8.81 s vs. 11.57 s) in seizure onset detection, and higher specificity (97.31% vs. 94.84%) and accuracy (96.85% vs. 95.42%). We can further reduce the latency of our algorithm to 3.74 s by allowing a slightly higher percentage of false alarms (2% specificity loss). Using only the top 10% of the electrodes ranked by our algorithm, we still maintain superior latency, sensitivity, and specificity compared to the other algorithms with all the electrodes. We finally demonstrate the suitability of our algorithm to deployment on low-cost embedded hardware platforms, thanks to its robustness to noise/artifacts affecting the signal, its low computational complexity, and the small memory-footprint on a RISC-V microcontroller.

Evaluation of artificial intelligence systems for assisting neurologists with fast and accurate annotations of scalp electroencephalography data

Background

Assistive automatic seizure detection can empower human annotators to shorten patient monitoring data review times. We present a proof-of-concept for a seizure detection system that is sensitive, automated, patient-specific, and tunable to maximise sensitivity while minimizing human annotation times. The system uses custom data preparation methods, deep learning analytics and electroencephalography (EEG) data.

Methods

Scalp EEG data of 365 patients containing 171,745 s ictal and 2,185,864 s interictal samples obtained from clinical monitoring systems were analysed as part of a crowdsourced artificial intelligence (AI) challenge. Participants were tasked to develop an ictal/interictal classifier with high sensitivity and low false alarm rates. We built a challenge platform that prevented participants from downloading or directly accessing the data while allowing crowdsourced model development.

Findings

The automatic detection system achieved tunable sensitivities between 75.00% and 91.60% allowing a reduction in the amount of raw EEG data to be reviewed by a human annotator by factors between 142x, and 22x respectively. The algorithm enables instantaneous reviewer-managed optimization of the balance between sensitivity and the amount of raw EEG data to be reviewed.

Interpretation

This study demonstrates the utility of deep learning for patient-specific seizure detection in EEG data. Furthermore, deep learning in combination with a human reviewer can provide the basis for an assistive data labelling system lowering the time of manual review while maintaining human expert annotation performance.

Funding

IBM employed all IBM Research authors. Temple University employed all Temple University authors. The Icahn School of Medicine at Mount Sinai employed Eren Ahsen. The corresponding authors Stefan Harrer and Gustavo Stolovitzky declare that they had full access to all the data in the study and that they had final responsibility for the decision to submit for publication.

Automated seizure detection using wearable devices: A clinical practice guideline of the International League Against Epilepsy and the International Federation of Clinical Neurophysiology

The objective of this clinical practice guideline (CPG) is to provide recommendations for healthcare personnel working with patients with epilepsy on the use of wearable devices for automated seizure detection in patients with epilepsy, in outpatient, ambulatory settings. The Working Group of the International League Against Epilepsy (ILAE) and the International Federation of Clinical Neurophysiology (IFCN) developed the CPG according to the methodology proposed by the ILAE Epilepsy Guidelines Working Group. We reviewed the published evidence using The Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) statement and evaluated the evidence and formulated the recommendations following the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system. We found high level of evidence for the accuracy of automated detection of generalized tonic-clonic seizures (GTCS) and focal-to-bilateral tonic-clonic seizures (FBTCS) and recommend the use of wearable automated seizure detection devices for selected patients when accurate detection of GTCS and FBTCS is recommended as a clinical adjunct. We also found a moderate level of evidence for seizure types without GTCS or FBTCS. However, it was uncertain whether the detected alarms resulted in meaningful clinical outcomes for the patients. We recommend using clinically validated devices for automated detection of GTCS and FBTCS, especially in unsupervised patients, where alarms can result in rapid intervention (weak/conditional recommendation). At present, we do not recommend clinical use of the currently available devices for other seizure types (weak/conditional recommendation). Further research and development are needed to improve the performance of automated seizure detection and to document their accuracy and clinical utility.

Automated seizure detection using wearable devices: A clinical practice guideline of the International League Against Epilepsy and the International Federation of Clinical Neurophysiology

The objective of this clinical practice guideline (CPG) is to provide recommendations for healthcare personnel working with patients with epilepsy, on the use of wearable devices for automated seizure detection in patients with epilepsy, in outpatient, ambulatory settings. The Working Group of the International League Against Epilepsy and the International Federation of Clinical Neurophysiology developed the CPG according to the methodology proposed by the ILAE Epilepsy Guidelines Working Group. We reviewed the published evidence using The Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) statement and evaluated the evidence and formulated the recommendations following the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system. We found high level of evidence for the accuracy of automated detection of generalized tonic-clonic seizures (GTCS) and focal-to-bilateral tonic-clonic seizures (FBTCS) and recommend use of wearable automated seizure detection devices for selected patients when accurate detection of GTCS and FBTCS is recommended as a clinical adjunct. We also found moderate level of evidence for seizure types without GTCs or FBTCs. However, it was uncertain whether the detected alarms resulted in meaningful clinical outcomes for the patients. We recommend using clinically validated devices for automated detection of GTCS and FBTCS, especially in unsupervised patients, where alarms can result in rapid intervention (weak/conditional recommendation). At present, we do not recommend clinical use of the currently available devices for other seizure types (weak/conditional recommendation). Further research and development are needed to improve the performance of automated seizure detection and to document their accuracy and clinical utility.

Automated detection of activity onset after postictal generalized EEG suppression

BACKGROUND

Sudden unexpected death in epilepsy (SUDEP) is a leading cause of premature death in patients with epilepsy. If timely assessment of SUDEP risk can be made, early interventions for optimized treatments might be provided. One of the biomarkers being investigated for SUDEP risk assessment is postictal generalized EEG suppression [postictal generalized EEG suppression (PGES)]. For example, prolonged PGES has been found to be associated with a higher risk for SUDEP. Accurate characterization of PGES requires correct identification of the end of PGES, which is often complicated due to signal noise and artifacts, and has been reported to be a difficult task even for trained clinical professionals. In this work we present a method for automatic detection of the end of PGES using multi-channel EEG recordings, thus enabling the downstream task of SUDEP risk assessment by PGES characterization.

METHODS

We address the detection of the end of PGES as a classification problem. Given a short EEG snippet, a trained model classifies whether it consists of the end of PGES or not. Scalp EEG recordings from a total of 134 patients with epilepsy are used for training a random forest based classification model. Various time-series based features are used to characterize the EEG signal for the classification task. The features that we have used are computationally inexpensive, making it suitable for real-time implementations and low-power solutions. The reference labels for classification are based on annotations by trained clinicians identifying the end of PGES in an EEG recording.

RESULTS

We evaluated our classification model on an independent test dataset from 34 epileptic patients and obtained an AUreceiver operating characteristic (ROC) (area under the curve) of 0.84. We found that inclusion of multiple EEG channels is important for better classification results, possibly owing to the generalized nature of PGES. Of among the channels included in our analysis, the central EEG channels were found to provide the best discriminative representation for the detection of the end of PGES.

CONCLUSION

Accurate detection of the end of PGES is important for PGES characterization and SUDEP risk assessment. In this work, we showed that it is feasible to automatically detect the end of PGES-otherwise difficult to detect due to EEG noise and artifacts-using time-series features derived from multi-channel EEG recordings. In future work, we will explore deep learning based models for improved detection and investigate the downstream task of PGES characterization for SUDEP risk assessment.

Machine learning approach to detect focal-onset seizures in the human anterior nucleus of the thalamus

OBJECTIVE

There is an unmet need to develop seizure detection algorithms from brain regions outside the epileptogenic cortex. The study aimed to demonstrate the feasibility of classifying seizures and interictal states from local field potentials (LFPs) recorded from the human thalamus- a subcortical region remote to the epileptogenic cortex. We tested the hypothesis that spectral and entropy-based features extracted from LFPs recorded from the anterior nucleus of the thalamus (ANT) can distinguish its state of ictal recruitment from other interictal states (including awake, sleep).

APPROACH

Two supervised machine learning tools (random forest and the random kitchen sink) were used to evaluate the performance of spectral (discrete wavelet transform-DWT), and time-domain (multiscale entropy-MSE) features in classifying seizures from interictal states in patients undergoing stereo EEG evaluation for epilepsy surgery. Under the supervision of IRB, field potentials were recorded from the ANT in consenting adults with drug-resistant temporal lobe epilepsy. Seizures were confirmed in the ANT using line-length and visual inspection. Wilcoxon rank-sum (WRS) method was used to test the differences in spectral patterns between seizure and interictal (awake and sleep) states.

MAIN RESULTS

79 seizures (10 patients) and 158 segments (approx. 4 hours) of interictal stereo EEG data were analyzed. The mean seizure detection latencies with line length in the ANT varied between seizure types (range 5-34 seconds). However, the DWT and MSE in the ANT showed significant changes for all seizure types within the first 20 seconds after seizure onset. The random forest (accuracy 93.9 % and false-positive 4.6%) and the random kitchen sink (accuracy 97.3% and false-positive 1.8%) classified seizures and interictal states.

SIGNIFICANCE

These results suggest that features extracted from the thalamic LFPs can be trained to detect seizures that can be used for monitoring seizure counts and for closed-loop seizure abortive interventions.

Accurate detection of spontaneous seizures using a generalized linear model with external validation

OBJECTIVE

Seizure detection is a major facet of electroencephalography (EEG) analysis in neurocritical care, epilepsy diagnosis and management, and the instantiation of novel therapies such as closed-loop stimulation or optogenetic control of seizures. It is also of increased importance in high-throughput, robust, and reproducible pre-clinical research. However, seizure detectors are not widely relied upon in either clinical or research settings due to limited validation. In this study, we create a high-performance seizure-detection approach, validated in multiple data sets, with the intention that such a system could be available to users for multiple purposes.

METHODS

We introduce a generalized linear model trained on 141 EEG signal features for classification of seizures in continuous EEG for two data sets. In the first (Focal Epilepsy) data set consisting of 16 rats with focal epilepsy, we collected 1012 spontaneous seizures over 3 months of 24/7 recording. We trained a generalized linear model on the 141 features representing 20 feature classes, including univariate and multivariate, linear and nonlinear, time, and frequency domains. We tested performance on multiple hold-out test data sets. We then used the trained model in a second (Multifocal Epilepsy) data set consisting of 96 rats with 2883 spontaneous multifocal seizures.

RESULTS

From the Focal Epilepsy data set, we built a pooled classifier with an Area Under the Receiver Operating Characteristic (AUROC) of 0.995 and leave-one-out classifiers with an AUROC of 0.962. We validated our method within the independently constructed Multifocal Epilepsy data set, resulting in a pooled AUROC of 0.963. We separately validated a model trained exclusively on the Focal Epilepsy data set and tested on the held-out Multifocal Epilepsy data set with an AUROC of 0.890. Latency to detection was under 5 seconds for over 80% of seizures and under 12 seconds for over 99% of seizures.

SIGNIFICANCE

This method achieves the highest performance published for seizure detection on multiple independent data sets. This method of seizure detection can be applied to automated EEG analysis pipelines as well as closed loop interventional approaches, and can be especially useful in the setting of research using animals in which there is an increased need for standardization and high-throughput analysis of large number of seizures.

Machine learning and wearable devices of the future

Machine learning (ML) is increasingly recognized as a useful tool in healthcare applications, including epilepsy. One of the most important applications of ML in epilepsy is seizure detection and prediction, using wearable devices (WDs). However, not all currently available algorithms implemented in WDs are using ML. In this review, we summarize the state of the art of using WDs and ML in epilepsy, and we outline future development in these domains. There is published evidence for reliable detection of epileptic seizures using implanted electroencephalography (EEG) electrodes and wearable, non-EEG devices. Application of ML using the data recorded with WDs from a large number of patients could change radically the way we diagnose and manage patients with epilepsy.

Seizure forecasting and cyclic control of seizures

Epilepsy is a unique neurologic condition characterized by recurrent seizures, where causes, underlying biomarkers, triggers, and patterns differ across individuals. The unpredictability of seizures can heighten fear and anxiety in people with epilepsy, making it difficult to take part in day-to-day activities. Epilepsy researchers have prioritized developing seizure prediction algorithms to combat episodic seizures for decades, but the utility and effectiveness of prediction algorithms has not been investigated thoroughly in clinical settings. In contrast, seizure forecasts, which theoretically provide the probability of a seizure at any time (as opposed to predicting the next seizure occurrence), may be more feasible. Many advances have been made over the past decade in the field of seizure forecasting, including improvements in algorithms as a result of machine learning and exploration of non-EEG-based measures of seizure susceptibility, such as physiological biomarkers, behavioral changes, environmental drivers, and cyclic seizure patterns. For example, recent work investigating periodicities in individual seizure patterns has determined that more than 90% of people have circadian rhythms in their seizures, and many also experience multiday, weekly, or longer cycles. Other potential indicators of seizure susceptibility include stress levels, heart rate, and sleep quality, all of which have the potential to be captured noninvasively over long time scales. There are many possible applications of a seizure-forecasting device, including improving quality of life for people with epilepsy, guiding treatment plans and medication titration, optimizing presurgical monitoring, and focusing scientific research. To realize this potential, it is vital to better understand the user requirements of a seizure-forecasting device, continue to advance forecasting algorithms, and design clear guidelines for prospective clinical trials of seizure forecasting.

Cross-Subject Seizure Detection in EEGs Using Deep Transfer Learning

Electroencephalography (EEG) plays an import role in monitoring the brain activities of patients with epilepsy and has been extensively used to diagnose epilepsy. Clinically reading tens or even hundreds of hours of EEG recordings is very time consuming. Therefore, automatic detection of seizure is of great importance. But the huge diversity of EEG signals belonging to different patients makes the task of seizure detection much challenging, for both human experts and automation methods. We propose three deep transfer convolutional neural networks (CNN) for automatic cross-subject seizure detection, based on VGG16, VGG19, and ResNet50, respectively. The original dataset is the CHB-MIT scalp EEG dataset. We use short time Fourier transform to generate time-frequency spectrum images as the input dataset, while positive samples are augmented due to the infrequent nature of seizure. The model parameters pretrained on ImageNet are transferred to our models. And the fine-tuned top layers, with an output layer of two neurons for binary classification (seizure or nonseizure), are trained from scratch. Then, the input dataset are randomly shuffled and divided into three partitions for training, validating, and testing the deep transfer CNNs, respectively. The average accuracies achieved by the deep transfer CNNs based on VGG16, VGG19, and ResNet50 are 97.75%, 98.26%, and 96.17% correspondingly. On those results of experiments, our method could prove to be an effective method for cross-subject seizure detection.

Wearable devices for seizure detection: Is it time to translate into our clinical practice?

With the exponential development of mobile health technologies over the past ten years, there has been a growing interest in the potential applications in the field of epilepsy, and specifically for seizure detection. Better detection of seizures is probably one of the best ways to improve patient safety. Overall, we are observing an exponential increase in the number of non-EEG based seizure detection systems and a progressive homogenization of their evaluation procedures. Most importantly, the properties of these devices for detection of tonic-clonic seizures are now very interesting, both in terms of sensitivity and in terms of false-alarm rates. Accordingly, we might expect that these be used in clinical practice in the near future, especially in patients at high risk of seizure-related injuries or sudden unexpected death in epilepsy (SUDEP).

Circadian and multiday seizure periodicities, and seizure clusters in canine epilepsy

Advances in ambulatory intracranial EEG devices have enabled objective analyses of circadian and multiday seizure periodicities, and seizure clusters in humans. This study characterizes circadian and multiday seizure periodicities, and seizure clusters in dogs with naturally occurring focal epilepsy, and considers the implications of an animal model for the study of seizure risk patterns, seizure forecasting and personalized treatment protocols. In this retrospective cohort study, 16 dogs were continuously monitored with ambulatory intracranial EEG devices designed for humans. Detailed medication records were kept for all dogs. Seizure periodicity was evaluated with circular statistics methods. Circular non-uniformity was assessed for circadian, 7-day and approximately monthly periods. The Rayleigh test was used to assess statistical significance, with correction for multiple comparisons. Seizure clusters were evaluated with Fano factor (index of dispersion) measurements, and compared to a Poisson distribution. Relationships between interseizure interval (ISI) and seizure duration were evaluated. Six dogs met the inclusion criteria of having at least 30 seizures and were monitored for an average of 65 weeks. Three dogs had seizures with circadian seizure periodicity, one dog had a 7-day periodicity, and two dogs had approximately monthly periodicity. Four dogs had seizure clusters and significantly elevated Fano factor values. There were subject-specific differences in the dynamics of ISI and seizure durations, both within and between lead and clustered seizure categories. Our findings show that seizure timing in dogs with naturally occurring epilepsy is not random, and that circadian and multiday seizure periodicities, and seizure clusters are common. Circadian, 7-day, and monthly seizure periodicities occur independent of antiseizure medication dosing, and these patterns likely reflect endogenous rhythms of seizure risk.

The role of devices in managing risk

Over the last few years, there has been significant expansion of wearable technologies and devices into the health sector, including for conditions such as epilepsy. Although there is significant potential to benefit patients, there is a paucity of well-conducted scientific research in order to inform patients and healthcare providers of the most appropriate technology. In addition to either directly or indirectly identifying seizure activity, the ideal device should improve quality of life and reduce the risk of sudden unexpected death in epilepsy (SUDEP). Devices typically monitor a number of parameters including electroencephalographic (EEG), cardiac, and respiratory patterns and can detect movement, changes in skin conductance, and muscle activity. Multimodal devices are emerging with improved seizure detection rates and reduced false positive alarms. While convulsive seizures are reliably identified by most unimodal and multimodal devices, seizures associated with no, or minimal, movement are frequently undetected. The vast majority of current devices detect but do not actively intervene. At best, therefore, they indicate the presence of seizure activity in order to accurately ascertain true seizure frequency or facilitate intervention by others, which may, nevertheless, impact the rate of SUDEP. Future devices are likely to both detect and intervene within an autonomous closed-loop system tailored to the individual and by self-learning from the analysis of patient-specific parameters. The formulation of standards for regulatory bodies to validate seizure detection devices is also of paramount importance in order to confidently ascertain the performance of a device; and this will be facilitated by the creation of a large, open database containing multimodal annotated data in order to test device algorithms. This paper is for the Special Issue: Prevent 21: SUDEP Summit - Time to Listen.

Crowdsourcing in health and medical research: a systematic review

Background

Crowdsourcing is used increasingly in health and medical research. Crowdsourcing is the process of aggregating crowd wisdom to solve a problem. The purpose of this systematic review is to summarize quantitative evidence on crowdsourcing to improve health.

Methods

We followed Cochrane systematic review guidance and systematically searched seven databases up to September 4th 2019. Studies were included if they reported on crowdsourcing and related to health or medicine. Studies were excluded if recruitment was the only use of crowdsourcing. We determined the level of evidence associated with review findings using the GRADE approach.

Results

We screened 3508 citations, accessed 362 articles, and included 188 studies. Ninety-six studies examined effectiveness, 127 examined feasibility, and 37 examined cost. The most common purposes were to evaluate surgical skills (17 studies), to create sexual health messages (seven studies), and to provide layperson cardio-pulmonary resuscitation (CPR) out-of-hospital (six studies). Seventeen observational studies used crowdsourcing to evaluate surgical skills, finding that crowdsourcing evaluation was as effective as expert evaluation (low quality). Four studies used a challenge contest to solicit human immunodeficiency virus (HIV) testing promotion materials and increase HIV testing rates (moderate quality), and two of the four studies found this approach saved money. Three studies suggested that an interactive technology system increased rates of layperson initiated CPR out-of-hospital (moderate quality). However, studies analyzing crowdsourcing to evaluate surgical skills and layperson-initiated CPR were only from high-income countries. Five studies examined crowdsourcing to inform artificial intelligence projects, most often related to annotation of medical data. Crowdsourcing was evaluated using different outcomes, limiting the extent to which studies could be pooled.

Conclusions

Crowdsourcing has been used to improve health in many settings. Although crowdsourcing is effective at improving behavioral outcomes, more research is needed to understand effects on clinical outcomes and costs. More research is needed on crowdsourcing as a tool to develop artificial intelligence systems in medicine.

Trial registration

PROSPERO: CRD42017052835. December 27, 2016.

Non-electroencephalography-based seizure detection

PURPOSE OF REVIEW

There is need for automated seizure detection using mobile or wearable devices, for objective seizure documentation and decreasing morbidity and mortality associated with seizures. Due to technological development, a high number of articles have addressed non-electroencephalography (EEG)-based seizure detection. However, the quality of study-design and reporting is extremely heterogeneous. We aimed at giving the reader a clear picture on the current state of seizure detection, describing the level of evidence behind the various devices.

RECENT FINDINGS

Fifteen studies of phase-2 or above, demonstrated that non-EEG-based devices detected generalized tonic-clonic seizures (GTCS) with high sensitivity (≥90%) and low false alarm rate (FAR) (down to 0.2/day). We found limited evidence for detection of motor seizures other than GTCS, mostly from subgroups in larger studies, targeting GTCS. There is little evidence for non-EEG-based detection of nonmotor seizures: sensitivity is low (19-74%) with extremely high FAR (50-216/day).

SUMMARY

Detection of GTCS is reliable and there are several, validated devices on the market. However, detection of other seizure types needs further research.