Fig 4 - uploaded by Giorgos Giannakakis
Content may be subject to copyright.
Architecture of the ECG seizure anticipation system comprising of 5 subsystems: (a) preprocessing, (b) HRV feature extraction, (c) normalization, (d) feature selection, (e) machine learning.
Source publication
Background and Objective: Heart rate variability parameters are studied by the research community as potential valuable indices for seizure
detection and anticipation. This paper investigates heart activity abnormalities during focal epileptic seizures in childhood.
Methods: Seizures affect both the sympathetic and parasympathetic system which is e...
Context in source publication
Context 1
... seizure anticipation procedure proposed in this study begins with the ECG preprocessing (Section 2.1), HRV parameters extraction (Section 2.2), baseline removal and pairwise transformation (Section 2.3), HRV features selection (Section 2.4). The architecture of the whole procedure is presented in Fig. 4 The pairwise transformation of the feature matrix with the selected retained features was used for the regression analysis. A partial least-squares regression model [40] The regressive response Yregr is a timeseries (see Fig. 7) which represents an "epileptic index" based on heart activity which increases with an imminent seizure. ...
Citations
... This system, tested on a group of 28 children, gained a sensitivity of 77.6% while keeping a False Alarm Ratio (FAR) of 2.56 events per night. Conversely, the authors of [52] compared different ML algorithms for detecting the pre-ictal transition in a sample of nine children. The best results were obtained with an SVM classifier with a multivariate input, yielding an accuracy of 77.1%. ...
The most critical burden for People with Epilepsy (PwE) is represented by seizures, the unpredictability of which severely impacts quality of life. The design of real-time warning systems that can detect or even predict ictal events would enhance seizure management, leading to high benefits for PwE and their caregivers. In the past, various research works highlighted that seizure onset is anticipated by significant changes in autonomic cardiac control, which can be assessed through heart rate variability (HRV). This manuscript conducted a scoping review of the literature analyzing HRV-based methods for detecting or predicting ictal events. An initial search on the PubMed database returned 402 papers, 72 of which met the inclusion criteria and were included in the review. These results suggest that seizure detection is more accurate in neonatal and pediatric patients due to more significant autonomic modifications during the ictal transitions. In addition, conventional metrics are often incapable of capturing cardiac autonomic variations and should be replaced with more advanced methodologies, considering non-linear HRV features and machine learning tools for processing them. Finally, studies investigating wearable systems for heart monitoring denoted how HRV constitutes an efficient biomarker for seizure detection in patients presenting significant alterations in autonomic cardiac control during ictal events.
... Variation of the time interval between a consistent point in time of each heartbeat (generally related to ventricular electrical activation), known as heart rate variability (HRV) (Rajendra Acharya et al., 2006), has been proven by numerous studies to be a useful indicator of physiological status (Ernst, 2017;Rajendra Acharya et al., 2006). Thanks to its non-invasive nature and strong connection to the autonomic nervous system (ANS) (Ernst, 2017;Rajendra Acharya et al., 2006), HRV has been adopted to study a wide range of diseases and clinical conditions, which include myocardial infarction (Buccelletti et al., 2009), sudden cardiac death (Melillo et al., 2013), diabetes (Kudat et al., 2006), renal failure (Ranpuria et al., 2008), sepsis (Bohanon et al., 2015), seizure (Giannakakis et al., 2019), and cancer (De Couck & Gidron, 2013). In addition, the emergence of wearable devices with heart monitoring capabilities has also allowed researchers to study the above-mentioned medical conditions in real-world settings (Perez et al., 2019), as well as in non-clinical applications such as sports (Dong, 2016), stress (Taelman et al., 2009), and sleep monitoring (Stein & Pu, 2012). ...
... Several approaches to identifying pre-ictal physiological alterations with the use of machine learning algorithms have been developed [15,[24][25][26][27][28], aiming to discriminate between pre-ictal and inter-ictal cardiac states based on their statistical footprints. Discriminative models based on statistical properties of the RR Interval (RRI) series of an ECG signal act as the base for most of the previously mentioned studies. ...
... Billeci et al. [26], 10 min/84.6%; and Giannakakis et al. [25], 21.8 s/77.1%. An added benefit from our proposed pipeline is that it is semi-supervised and immediately deployable without the need of a pre-trained classifier. ...
... It is important to clarify that our findings indicate a pre-ictal increase in HRV novelty that is linked to ANS abnormalities; this is not, however, the only possible source of novelty in an NNI series. As has already been discussed by Giannakakis and colleagues [25], heart rate is a highly variable modality, altering in relation to physical activity, psychological condition, etc., which may induce false alarms if not provided as part of the reference interval. Specifically, in [27], an experimental case is described in which a patient's heart rate was significantly disrupted by the patient performing a trivial activity, greatly altering system performance. ...
Epilepsy is one of the most common brain diseases, characterized by frequent recurrent seizures or “ictal” states. A patient experiences uncontrollable muscular contractions, inducing loss of mobility and balance, which may result in injury or even death during these ictal states. Extensive investigation is vital to establish a systematic approach for predicting and informing patients about oncoming seizures ahead of time. Most methodologies developed are focused on the detection of abnormalities using mostly electroencephalogram (EEG) recordings. In this regard, research has indicated that certain pre-ictal alterations in the Autonomic Nervous System (ANS) can be detected in patient electrocardiogram (ECG) signals. The latter could potentially provide the basis for a robust seizure prediction approach. The recently proposed ECG-based seizure warning systems utilize machine learning models to classify a patient’s condition. Such approaches require the incorporation of large, diverse, and thoroughly annotated ECG datasets, limiting their application potential. In this work, we investigate anomaly detection models in a patient-specific context with low supervision requirements. Specifically, we consider One-Class SVM (OCSVM), Minimum Covariance Determinant (MCD) Estimator, and Local Outlier Factor (LOF) models to quantify the novelty or abnormality of pre-ictal short-term (2–3 min) Heart Rate Variability (HRV) features of patients, trained on a reference interval considered to contain stable heart rate as the only form of supervision. Our models are evaluated against labels that were either hand-picked or automatically generated (weak labels) by a two-phase clustering procedure for samples of the “Post-Ictal Heart Rate Oscillations in Partial Epilepsy” (PIHROPE) dataset recorded by the Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, achieving detection in 9 out of 10 cases, with average AUCs of over 93% across all models and warning times ranging from 6 to 30 min prior to seizure. The proposed anomaly detection and monitoring approach can potentially pave the way for early detection and warning of seizure incidents based on body sensor inputs.
... The ectopic heartbeats (irregular heartbeats deviated from normal) were also detected and excluded by adopting the HRV signal approach (percentage change of 70% over the averaged previous 5 heartbeats). The whole preprocessing procedure is described in [21]. ...
Background:
Preterm neonates show decreased HRV compared to those at full-term. We compared HRV metrics between preterm and full-term neonates in transfer periods from neonate rest state to neonate-parent interaction, and vice versa.
Methods:
Short-term recordings of the HRV parameters (time and frequency-domain indices and non-linear measurements) of 28 premature healthy neonates were compared with the metrics of 18 full-term neonates. HRV recordings were performed at home at term-equivalent age and HRV metrics were compared between the following transfer periods: from first rest state of the neonate (TI1) to a period in which the neonate interacted with the first parent (TI2), from TI2 to a second neonate rest state (TI3), and from TI3 to a period of neonate interaction with the second parent (TI4).
Results:
For the whole HRV recording period, PNN50, NN50 and HF (%) was lower for preterm neonates compared to full-terms. These findings support the reduced parasympathetic activity of preterm compared to full-term neonates. The results of comparisons between transfer period simply a common coactivation of SNS and PNS systems for both full and pre-term neonates.
Conclusions:
Spontaneous interaction with the parent may reinforce both full and pre-term neonates' ANS maturation.
... Many research groups have found that the values of some of these HRV features significantly increase or decrease when a seizure is about to start or has started [136][137][138][139][140][141][142][143][144], which makes them valuable biomarkers for seizure detection and seizure prediction. These variations are measured against the baseline, i.e., the interictal period. ...
Around the world, about 50 million people suffer from epilepsy, a disease characterized by recurrent and unprovoked seizures of abrupt cerebral activity. Epileptic seizures can present a serious danger to the daily life of patients and the ones surrounding them. Predicting epileptic seizures before they begin, therefore anticipating them, would allow patients to prepare and to go to a safe location, which represents an enormous improvement in their quality of life. Some epileptic seizures elicit abrupt changes in the autonomic nervous system (ANS) during the seizure outbreak and, sometimes, in the minutes that precede it. The ANS is responsible for cardiovascular and sudomotor control, so these changes can be identified by acquiring biosignals from the heart, muscles and sweat glands. This knowledge has opened doors for the development of continuous monitoring wearable devices to detect – some even to predict – epileptic seizures based solely on those ANS changes. In this project, through an extensive literature review of the methods published so far, the requirements for the development of a system for online seizure detection and prediction in ambulatory environments have been gathered, and a system architecture to be implemented in wearable devices has been proposed.
... Therefore, any additional physiological signal that can support the prediction of a seizure is valuable. In this task, attempts are made to use the ECG signal [10]. Numerous methods of quantifying the heart rate variability have been proposed [11][12][13]. ...
CSI is based on the quantitative analyses of Poincare plot features. Several studies indicate that it is possible to use CSI to detect and predict seizures [14-17]. For CSI calculation, the method proposed by Toichi et al. in [18] is most often used. In this article, the symbolic name "Toichi" is used to refer to the algorithm described in this work. To determine CSI, first R-waves must be detected in the ECG signal. For this purpose, the well-known Pan-Tompkins algorithm [19] or another method can be used. Next, the RR intervals should be analysed using the Poincare plot [20]. The purpose of the article is to use a 1D-convolutional neu-ral network [21-23] to calculate the CSI values based on the registered ECG signal. For a trained neural network, there is no need to perform R-wave detection [24] and Poincare plot analysis. It was assumed that deep learning, with many training examples from many patients, would allow the network to work effectively. A 1D-convolutional neural network was used to estimate the CSI. Then it was checked whether the CSI was reliable for the prediction of epileptic seizures. It was also examined how the loss of part of the recorded signal or artifacts affected the calculation of the CSI using both the standard Toichi algorithm and the proposed method. The presented research is an extension of the study described in the conference paper [25]. 2. Materials Video EEG (vEEG) with simultaneous ECG was recorded in patients with refractory temporal epilepsy for long-term monitoring of epileptic seizures during presurgical evaluation. The signals were acquired at the Medical University of Warsaw.
... Previous studies of autonomic modulation in children with epilepsy published different results, most of which indicate the impairment of CANS regulation considering HRV measurements in the time and frequency domain (14,15,(19)(20)(21)(22)(23)(24)(25)(26). In addition to 24-h long-term analysis, recent studies also focused on ictal or peri-ictal characters with the HRV method to investigate heart activity abnormalities or detect seizures (27)(28)(29)(30). The inconsistency in the results, however, was probably owing to different designs of the experiments in sample size, epilepsy type, recording time, and analysis detail. ...
... Because the EEG data is considered as the gold standard of identification of the seizure episodes, there might remain undetected seizure episodes in our pre-processed ECG data. They might impact our results because some studies have published the effects of seizure episodes on heart rate and HRV (27)(28)(29)(30). Further studies are needed to explore the impact of seizures on the CANS of pre-school children. ...
Objective: Intractable epilepsy and uncontrolled seizures could affect cardiac function and the autonomic nerve system with a negative impact on children's growth. The aim of this study was to investigate the variability and complexity of cardiac autonomic function in pre-school children with pediatric intractable epilepsy (PIE).
Methods: Twenty four-hour Holter electrocardiograms (ECGs) from 93 patients and 46 healthy control subjects aged 3–6 years were analyzed by the methods of traditional heart rate variability (HRV), multiscale entropy (MSE), and Kurths–Wessel symbolization entropy (KWSE). Receiver operating characteristic (ROC) curve analysis was used to estimate the overall discrimination ability. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) models were also analyzed.
Results: Pre-school children with PIE had significantly lower HRV measurements than healthy controls in time (Mean_RR, SDRR, RMSSD, pNN50) and frequency (VLF, LF, HF, LF/HF, TP) domains. For the MSE analysis, area 1_5 in awake state was lower, and areas 6_15 and 6_20 in sleep state were higher in PIE with a significant statistical difference. KWSE in the PIE group was also inferior to that in healthy controls. In ROC curve analysis, pNN50 had the greatest discriminatory power for PIE. Based on both NRI and IDI models, the combination of MSE indices (wake: area1_5 and sleep: area6_20) and KWSE (m = 2, τ = 1, α = 0.16) with traditional HRV measures had greater discriminatory power than any of the single HRV measures.
Significance: Impaired HRV and complexity were found in pre-school children with PIE. HRV, MSE, and KWSE could discriminate patients with PIE from subjects with normal cardiac complexity. These findings suggested that the MSE and KWSE methods may be helpful for assessing and understanding heart rate dynamics in younger children with epilepsy.
... In human medicine, HRV analysis has been utilised across a wide variety of fields, from detecting foetal distress, predicting heart failure following myocardial infarction, predicting the onset of seizures, assessing the effect of overtraining in athletes or to encourage lifestyle modification (Woo et al., 1992;Perini and Veicsteinas, 2003;Reed et al., 2005;Perkiomaki et al., 2014;Disertori et al., 2016;Singh et al., 2018;Giannakakis et al., 2019;Li et al., 2019). ...
... The use of HRV analyses in humans has expanded far beyond that limited to just the field of cardiology. In particular, sports medicine and many areas of lifestyle management have adapted HRV techniques to detect overtraining in elite athletes, predict the development of acute mountain sickness in mountaineers or alert a person with epilepsy to an imminent seizure, among many other uses (Mellor et al., 2018;Singh et al., 2018;Giannakakis et al., 2019). The second-degree AV blocks have been removed by the filter, including approximately 0.7% of the RR intervals in the data set. ...
Heart rate variability (HRV) analysis has been performed on ECG-derived data sets for more than 170 years but is currently undergoing a rapid evolution, thanks to the expansion of the human and veterinary medical technology sector. Traditional HRV analysis was initially performed to identify changes in vago-sympathetic balance, while the most recent focus has expanded to include the use of complex computer algorithms, neural networks and machine learning technology to identify cardiac arrhythmias, particularly atrial fibrillation (AF). Some of these techniques have recently been translated for use in the field of equine cardiology, with particular focus on improving the diagnosis of arrhythmias both at rest and during exercise. This review focuses on understanding the basic HRV variables and important factors to consider when collecting data for use in HRV analysis. In addition, the use of HRV analysis for the diagnosis of arrhythmias is discussed from human, small animal and equine perspectives. Finally, the future of HRV analysis is briefly introduced, including an overview of future developments in this rapidly expanding and exciting field.
... In the last decade, several researchers worldwide have presented diverse methods or methodologies based on electroencephalogram (EEG) signals for predicting an epileptic seizure [7-16]; however, despite obtaining promising results by using EEG signals, diverse studies presented in recent years indicate that the autonomic nervous system can produce alterations also in the electrocardiogram (ECG) signals [17][18][19][20], becoming a promising information source that can be used for predicting an epileptic seizure [21][22][23][24][25]. For example, Popov et al. [21] integrated 112 features (i.e., histogram characteristics, spectral analysis, polynomial approximation coefficients, among others) with a support vector machine (SVM) for predicting an epileptic seizure using the heart rate variability (HRV) of an ECG signal. ...
... The obtained results show an accuracy of 74.6% for predicting an epileptic seizure 10 min before the onset. Giannakakis et al. [25] integrated 18 diverse nonlinear methods: time-domain (envelope, the mean and standard deviation of RR intervals, the mean heart rate, among others) and frequency-domain (power in a low and high frequency band, frequency of the high and low frequency band peak, among others) with an SVM for predicting an epileptic seizure by using the HRV of ECG signals. The authors reported an accuracy of 77.1% for predicting an epileptic seizure 21.8 s before the onset. ...
... It should be noticed that 100% accuracy is obtained, which is a reasonable improvement over the presented results of similar works [21][22][23][24][25]; moreover, the proposal achieves a 15 min interval for the seizure prediction, being a good time window for allowing to take remedial actions to avoid severe damage to patient integrity. As noted by Vargas-Lopez et al. [28], it is desirable to achieve a methodology capable of obtaining a 100% accuracy as this will indicate that theoretically, any patient prone to suffer an episode can have a timely alert; in this sense, the proposal achieves this desired scenario. ...
Epilepsy is a brain disorder that affects about 50 million persons around the world and is characterized by generating recurrent seizures, which can put patients in permanent because of falls, drowning, burns, and prolonged seizures that they can suffer. Hence, it is of vital importance to propose a methodology with the capability of predicting a seizure with several minutes before the onset, allowing that the patients take their precautions against injuries. In this regard, a methodology based on the wavelet packet transform (WPT), statistical time features (STFs), and a decision tree classifier (DTC) for predicting an epileptic seizure using electrocardiogram (ECG) signals is presented. Seventeen STFs were analyzed to measure changes in the properties of ECG signals and find characteristics capable of differentiating between healthy and 15 min prior to seizure signals. The effectiveness of the proposed methodology for predicting an epileptic event is demonstrated using a database of seven patients with 10 epileptic seizures, which was provided by the Massachusetts Institute of Technology–Beth Israel Hospital (MIT–BIH). The results show that the proposed methodology is capable of predicting an epileptic seizure 15 min before with an accuracy of 100%. Our results suggest that the use of STFs at frequency bands related to heart activity to find parameters for the prediction of epileptic seizures is suitable.
... In Ref. 2, a deep learning model able to detect medically interpretable information in relevant images from a volume to classify diabetesrelated retinal diseases was introduced. The authors in Ref. 3 investigated heart activity abnormalities during focal epileptic seizures in childhood, in order to anticipate focal seizures. The authors in Ref. 4 developed a fully automatic and realtime ventricular heartbeat classi¯er based on a single electrocardiogram (ECG) that can be especially useful for wearable technologies that provide continuous and longterm monitoring. ...
Chaotic time series prediction can be performed by applying different architectures of artificial neural networks (ANNs) that can be implemented on field-programmable gate arrays (FPGAs). However, the main challenges are the reduction of hardware resources to develop faster ANNs and the prediction capabilities for large horizons. In this manner, the contribution is devoted to introduce pipeline architectures in which some registers are placed between combinational blocks to divide the logic into shorter stages that can run with a faster clock. The cases of study are the multilayer perceptron (MLP), nonlinear autoregressive with exogenous input (NARX), and echo state network (ESN). In addition, another contribution is devoted to introduce the application of the decimation technique to extend the prediction horizon of the ANNs from 12 to 600-steps-ahead. The prediction capabilities of the MLP, NARX and ESN are compared by using eight chaotic time series with different maximum Lyapunov exponents. The pipeline FPGA-based implementations show that the ESN with a reservoir of at least 30 neurons guarantees a large prediction horizon of 600-steps-ahead. Another important advantage of the ESN is that its FPGA-based implementation can be performed by reusing one neuron, thus requiring the lowest quantity of hardware resources.
