Robust Epileptic Seizure Detection on Wearable Systems with Reduced False-Alarm Rate

Reducing False Alarms in Wearable Seizure Detection With EEGformer: A Compact Transformer Model for MCUs

The long-term, continuous analysis of electroencephalography (EEG) signals on wearable devices to automatically detect seizures in epileptic patients is a high-potential application field for deep neural networks, and specifically for transformers, which are highly suited for end-to-end time series processing without handcrafted feature extraction. In this work, we propose a small-scale transformer detector, the EEGformer, compatible with unobtrusive acquisition setups that use only the temporal channels. EEGformer is the result of a hardware-oriented design exploration, aiming for efficient execution on tiny low-power micro-controller units (MCUs) and low latency and false alarm rate to increase patient and caregiver acceptance.Tests conducted on the CHB-MIT dataset show a 20% reduction of the onset detection latency with respect to the state-of-the-art model for temporal acquisition, with a competitive 73% seizure detection probability and 0.15 false-positive-per-hour (FP/h). Further investigations on a novel and challenging scalp EEG dataset result in the successful detection of 88% of the annotated seizure events, with 0.45 FP/h.We evaluate the deployment of the EEGformer on three commercial low-power computing platforms: the single-core Apollo4 MCU and the GAP8 and GAP9 parallel MCUs. The most efficient implementation (on GAP9) results in as low as 13.7 ms and 0.31 mJ per inference, demonstrating the feasibility of deploying the EEGformer on wearable seizure detection systems with reduced channel count and multi-day battery duration.

Minimizing artifact-induced false-alarms for seizure detection in wearable EEG devices with gradient-boosted tree classifiers

Electroencephalography (EEG) is widely used to monitor epileptic seizures, and standard clinical practice consists of monitoring patients in dedicated epilepsy monitoring units via video surveillance and cumbersome EEG caps. Such a setting is not compatible with long-term tracking under typical living conditions, thereby motivating the development of unobtrusive wearable solutions. However, wearable EEG devices present the challenges of fewer channels, restricted computational capabilities, and lower signal-to-noise ratio. Moreover, artifacts presenting morphological similarities to seizures act as major noise sources and can be misinterpreted as seizures. This paper presents a combined seizure and artifacts detection framework targeting wearable EEG devices based on Gradient Boosted Trees. The seizure detector achieves nearly zero false alarms with average sensitivity values of [Formula: see text] for 182 seizures from the CHB-MIT dataset and [Formula: see text] for 25 seizures from the private dataset with no preliminary artifact detection or removal. The artifact detector achieves a state-of-the-art accuracy of [Formula: see text] (on the TUH-EEG Artifact Corpus dataset). Integrating artifact and seizure detection significantly reduces false alarms-up to [Formula: see text] compared to standalone seizure detection. Optimized for a Parallel Ultra-Low Power platform, these algorithms enable extended monitoring with a battery lifespan reaching 300 h. These findings highlight the benefits of integrating artifact detection in wearable epilepsy monitoring devices to limit the number of false positives.

Automatic Detection and Classification of Epileptic Seizures from EEG Data: Finding Optimal Acquisition Settings and Testing Interpretable Machine Learning Approach

Deep learning (DL) is emerging as a successful technique for automatic detection and differentiation of spontaneous seizures that may otherwise be missed or misclassified. Herein, we propose a system architecture based on top-performing DL models for binary and multigroup classifications with the non-overlapping window technique, which we tested on the TUSZ dataset. The system accurately detects seizure episodes (87.7% Sn, 91.16% Sp) and carefully distinguishes eight seizure types (95-100% Acc). An increase in EEG sampling rate from 50 to 250 Hz boosted model performance: the precision of seizure detection rose by 5%, and seizure differentiation by 7%. A low sampling rate is a reasonable solution for training reliable models with EEG data. Decreasing the number of EEG electrodes from 21 to 8 did not affect seizure detection but worsened seizure differentiation significantly: 98.24 ± 0.17 vs. 85.14 ± 3.14% recall. In detecting epileptic episodes, all electrodes provided equally informative input, but in seizure differentiation, their informative value varied. We improved model explainability with interpretable ML. Activation maximization highlighted the presence of EEG patterns specific to eight seizure types. Cortical projection of epileptic sources depicted differences between generalized and focal seizures. Interpretable ML techniques confirmed that our system recognizes biologically meaningful features as indicators of epileptic activity in EEG.

Importance of methodological choices in data manipulation for validating epileptic seizure detection models

Epilepsy is a chronic neurological disorder that affects a significant portion of the human population and imposes serious risks in the daily life. Despite advances in machine learning and IoT, small, non-stigmatizing wearable devices for continuous monitoring and detection in outpatient environments are not yet widely available. Part of the reason is the complexity of epilepsy itself, including highly imbalanced data, multimodal nature, and very subject-specific signatures. However, another problem is the heterogeneity of methodological approaches in research, leading to slower progress, difficulty in comparing results, and low reproducibility. Therefore, this article identifies a wide range of methodological decisions that must be made and reported when training and evaluating the performance of epilepsy detection systems. We characterize the influence of individual choices using a typical ensemble random-forest model and the publicly available CHB-MIT database, providing a broader picture of each decision and giving good-practice recommendations, based on our experience, where possible.

A Shallow Autoencoder Framework for Epileptic Seizure Detection in EEG Signals

This paper presents a trainable hybrid approach involving a shallow autoencoder (AE) and a conventional classifier for epileptic seizure detection. The signal segments of a channel of electroencephalogram (EEG) (EEG epochs) are classified as epileptic and non-epileptic by employing its encoded AE representation as a feature vector. Analysis on a single channel-basis and the low computational complexity of the algorithm allow its use in body sensor networks and wearable devices using one or few EEG channels for wearing comfort. This enables the extended diagnosis and monitoring of epileptic patients at home. The encoded representation of EEG signal segments is obtained based on training the shallow AE to minimize the signal reconstruction error. Extensive experimentation with classifiers has led us to propose two versions of our hybrid method: (a) one yielding the best classification performance compared to the reported methods using the k-nearest neighbor (kNN) classifier and (b) the second with a hardware-friendly architecture and yet with the best classification performance compared to other reported methods in this category using a support-vector machine (SVM) classifier. The algorithm is evaluated on the Children's Hospital Boston, Massachusetts Institute of Technology (CHB-MIT), and University of Bonn EEG datasets. The proposed method achieves 98.85% accuracy, 99.29% sensitivity, and 98.86% specificity on the CHB-MIT dataset using the kNN classifier. The best figures using the SVM classifier for accuracy, sensitivity, and specificity are 99.19%, 96.10%, and 99.19%, respectively. Our experiments establish the superiority of using an AE approach with a shallow architecture to generate a low-dimensionality yet effective EEG signal representation capable of high-performance abnormal seizure activity detection at a single-channel EEG level and with a fine granularity of 1 s EEG epochs.

Approximate zero-crossing: a new interpretable, highly discriminative and low-complexity feature for EEG and iEEG seizure detection

Objective. Long-term monitoring of people with epilepsy based on electroencephalography (EEG) and intracranial EEG (iEEG) has the potential to deliver key clinical information for personalised epilepsy treatment. More specifically, in outpatient settings, the available solutions are not satisfactory either due to poor classification performance or high complexity to be executed in resource-constrained devices (e.g. wearable systems). Therefore, we hypothesize that obtaining high discriminative features is the main avenue to improve low-complexity seizure-detection algorithms.Approach. Inspired by how neurologists recognize ictal EEG data, and to tackle this problem by targeting resource-constrained wearable devices, we introduce a new interpretable and highly discriminative feature for EEG and iEEG, namely approximate zero-crossing (AZC). We obtain AZC by applying a polygonal approximation to mimic how our brain selects prominent patterns among noisy data and then using a zero-crossing count as a measure of the dominating frequency. By employing Kullback-Leiber divergence, leveraging CHB-MIT and SWEC-ETHZ iEEG datasets, we compare the AZC discriminative power against a set of 56 classical literature features (CLF). Moreover, we assess the performances of a low-complexity seizure detection method using only AZC features versus employing the CLF set.Main results. Three AZC features obtained with different approximation thresholds are among the five with the highest median discriminative power. Moreover, seizure classification based on only AZC features outperforms an equivalent CLF-based method. The former detects 102 and 194 seizures, against 99 and 161 for the latter (CHB-MIT and SWEC-ETHZ, respectively). Moreover, the AZC-based method keeps a similar false-alarm rate (i.e. an average of 2.1 and 1.0, against 2.0 and 0.5, per day).Significance. We propose a new feature and demonstrate its capability in seizure classification for both scalp and intracranial EEG. We envision the use of such a feature to improve outpatient monitoring with resource-constrained devices.

Exploration of Hyperdimensional Computing Strategies for Enhanced Learning on Epileptic Seizure Detection

Wearable and unobtrusive monitoring and prediction of epileptic seizures has the potential to significantly increase the life quality of patients, but is still an unreached goal due to challenges of real-time detection and wearable devices design. Hyperdimensional (HD) computing has evolved in recent years as a new promising machine learning approach, especially when talking about wearable applications. But in the case of epilepsy detection, standard HD computing is not performing at the level of other state-of-the-art algorithms. This could be due to the inherent complexity of the seizures and their signatures in different biosignals, such as the electroencephalogram (EEG), the highly personalized nature, and the disbalance of seizure and non-seizure instances. In the literature, different strategies for improved learning of HD computing have been proposed, such as iterative (multi-pass) learning, multi-centroid learning and learning with sample weight ("OnlineHD"). Yet, most of them have not been tested on the challenging task of epileptic seizure detection, and it stays unclear whether they can increase the HD computing performance to the level of the current state-of-the-art algorithms for wearable devices, such as random forests. Thus, in this paper, we implement different learning strategies and assess their performance on an individual basis, or in combination, regarding detection performance and memory and computational requirements. Results show that the best-performing algorithm, which is a combination of multi-centroid and multi-pass, can indeed reach the performance of the random forest model on a highly unbalanced dataset imitating a real-life epileptic seizure detection application.

Making remote measurement technology work in multiple sclerosis, epilepsy and depression: survey of healthcare professionals

Background

Epilepsy, multiple sclerosis (MS) and depression are long term, central nervous system disorders which have a significant impact on everyday life. Evaluating symptoms of these conditions is problematic and typically involves repeated visits to a clinic. Remote measurement technology (RMT), consisting of smartphone apps and wearables, may offer a way to improve upon existing methods of managing these conditions. The present study aimed to establish the practical requirements that would enable clinical integration of data from patients’ RMT, according to healthcare professionals.

Methods

This paper reports findings from an online survey of 1006 healthcare professionals currently working in the care of people with epilepsy, MS or depression. The survey included questions on types of data considered useful, how often data should be collected, the value of RMT data, preferred methods of accessing the data, benefits and challenges to RMT implementation, impact of RMT data on clinical practice, and requirement for technical support. The survey was presented on the JISC online surveys platform.

Results

Among this sample of 1006 healthcare professionals, respondents were positive about the benefits of RMT, with 73.2% indicating their service would be likely or highly likely to benefit from the implementation of RMT in patient care plans. The data from patients’ RMT devices should be made available to all nursing and medical team members and could be reviewed between consultations where flagged by the system. However, results suggest it is also likely that RMT data would be reviewed in preparation for and during a consultation with a patient. Time to review information is likely to be one of the greatest barriers to successful implementation of RMT in clinical practice.

Conclusions

While further work would be required to quantify the benefits of RMT in clinical practice, the findings from this survey suggest that a wide array of clinical team members treating epilepsy, MS and depression would find benefit from RMT data in the care of their patients. Findings presented could inform the implementation of RMT and other digital interventions in the clinical management of a range of neurological and mental health conditions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12911-022-01856-z.

Multi-Centroid Hyperdimensional Computing Approach for Epileptic Seizure Detection

Long-term monitoring of patients with epilepsy presents a challenging problem from the engineering perspective of real-time detection and wearable devices design. It requires new solutions that allow continuous unobstructed monitoring and reliable detection and prediction of seizures. A high variability in the electroencephalogram (EEG) patterns exists among people, brain states, and time instances during seizures, but also during non-seizure periods. This makes epileptic seizure detection very challenging, especially if data is grouped under only seizure (ictal) and non-seizure (inter-ictal) labels. Hyperdimensional (HD) computing, a novel machine learning approach, comes in as a promising tool. However, it has certain limitations when the data shows a high intra-class variability. Therefore, in this work, we propose a novel semi-supervised learning approach based on a multi-centroid HD computing. The multi-centroid approach allows to have several prototype vectors representing seizure and non-seizure states, which leads to significantly improved performance when compared to a simple single-centroid HD model. Further, real-life data imbalance poses an additional challenge and the performance reported on balanced subsets of data is likely to be overestimated. Thus, we test our multi-centroid approach with three different dataset balancing scenarios, showing that performance improvement is higher for the less balanced dataset. More specifically, up to 14% improvement is achieved on an unbalanced test set with 10 times more non-seizure than seizure data. At the same time, the total number of sub-classes is not significantly increased compared to the balanced dataset. Thus, the proposed multi-centroid approach can be an important element in achieving a high performance of epilepsy detection with real-life data balance or during online learning, where seizures are infrequent.

Systematic Assessment of Hyperdimensional Computing for Epileptic Seizure Detection

Hyperdimensional computing is a promising novel paradigm for low-power embedded machine learning. It has been applied on different biomedical applications, and particularly on epileptic seizure detection. Unfortunately, due to differences in data preparation, segmentation, encoding strategies, and performance metrics, results are hard to compare, which makes building upon that knowledge difficult. Thus, the main goal of this work is to perform a systematic assessment of the HD computing framework for the detection of epileptic seizures, comparing different feature approaches mapped to HD vectors. More precisely, we test two previously implemented features as well as several novel approaches with HD computing on epileptic seizure detection. We evaluate them in a comparable way, i.e., with the same preprocessing setup and with identical performance measures. We use two different datasets in order to assess the generalizability of our conclusions. The systematic assessment involved three primary aspects relevant for potential wearable implementations: 1) detection performance, 2) memory requirements, and 3) computational complexity. Our analysis shows a significant difference in detection performance between approaches, but also that the ones with the highest performance might not be ideal for wearable applications due to their high memory or computational requirements. Furthermore, we evaluate a post-processing strategy to adjust the predictions to the dynamics of epileptic seizures, showing that performance is significantly improved in all the approaches and also that after post-processing, differences in performance are much smaller between approaches.

Classifying Single Channel Epileptic EEG data based on Sparse Representation using Shallow Autoencoder

Patient independent epileptic seizure detection algorithm for scalp electroencephalogram (EEG) data is pro- posed in this paper. Principal motivation of this work is to integrate neural and conventional machine learning methods to develop a classification system which can advance the current wearable health systems in terms of computational complexity and accuracy. Being based on processing a single channel EEG processing, the approach is suitable for usage with small wireless sensors. A shallow autoencoder model is utilized for sparse representation of the EEG signal followed by k-nearest neighbor (kNN) classifier to categorize the data as epileptic or non-epileptic. Using a single EEG channel an optimum sparsity level is explored in the encoded sample. Attaining an accuracy, sensitivity and specificity of 98.85%, 99.29% and 98.86% respectively, for CHB-MIT scalp EEG database, proposed classification method outperforms state of- the-art seizure detection methodologies. Experiments has shown that this performance was possible by using a sparsity level of 4 in the auto-encoder. Furthermore, use of shallow learning instead of deep learning approach for generation of sparse but effective representation is computationally lighter than many other feature extraction and preprocessing methods.

MBioTracker: Multimodal Self-Aware Bio-Monitoring Wearable System for Online Workload Detection

Cognitive workload affects operators' performance principally in high-risk or time-demanding situations and when multitasking is required. An online cognitive workload monitoring system can provide valuable inputs to decision-making instances, such as the operator's state of mind and resulting performance. Therefore, it can allow potential adaptive support to the operator. This work presents a new design of a wearable embedded system for online cognitive workload monitoring. This new wearable system consists of, on the hardware side, a multi-channel physiological signals acquisition (respiration cycles, heart rate, skin temperature, and pulse waveform) and a low-power processing platform. Further, on the software side, our wearable embedded system includes a novel energy-aware bio-signal processing algorithm. We also use the concept of application self-awareness to enable energy-scalable embedded machine learning algorithms and methods for online subjects' cognitive workload monitoring. Our results show that this new wearable system can continuously monitor multiple bio-signals, compute their key features, and provide reliable detection of high and low cognitive workload levels with a time resolution of 1 minute and a battery lifetime of 14.58 h in our experimental conditions. It achieves a detection accuracy of 76.6% (2.6% lower than analogous offline computer-based analysis) with a sensitivity of 77.04% and a specificity of 81.75%, on a simulated drone rescue mission task. Moreover, by applying our self-aware monitoring to exploit different energy-scalable modes, we can increase battery lifetime by 51.6% (up to 22.11 hours) while incurring an insignificant accuracy loss of 1.07%.

Real-Time EEG-Based Cognitive Workload Monitoring on Wearable Devices

OBJECTIVE

Cognitive workload monitoring (CWM) can enhance human-machine interaction by supporting task execution assistance considering the operators cognitive state. Therefore, we propose a machine learning design methodology and a data processing strategy to enable CWM on resource-constrained wearable devices.

METHODS

Our CWM solution is built upon edge computing on a simple wearable system, with only four peripheral channels of electroencephalography (EEG). We assess our solution on experimental data from 24 volunteers. Moreover, to overcome the system's memory constraints, we adopt an optimization strategy for model size reduction and a multi-batch data processing scheme for optimizing RAM memory footprint. Finally, we implement our data processing strategy on a state-of-the-art wearable platform and assess its execution and system battery life.

RESULTS

We achieve an accuracy of 74.5% and a 74.0% geometric mean between sensitivity and specificity for CWM classification on unseen data. Besides, the proposed model optimization strategy generates a 27.5x smaller model compared to the one generated with default parameters, and the multi-batch data processing scheme reduces RAM memory footprint by 14x compared to a single batch data processing. Finally, our algorithm uses only 1.28% of the available processing time, thus, allowing our system to achieve 28.5 hours of battery life.

CONCLUSION

We provide a reliable and optimized CWM solution using wearable devices, enabling more than a day of operation on a single battery charge.

SIGNIFICANCE

The proposed methodology enables real-time data processing on resource-constrained devices and supports real-time wearable monitoring based on EEG for applications as CWM in human-machine interaction.