Publications
710 results found
Żyliński M, Nassibi A, Mandic DP, 2023, Design and Implementation of an Atrial Fibrillation Detection Algorithm on the ARM Cortex-M4 Microcontroller., Sensors (Basel), Vol: 23
At present, a medium-level microcontroller is capable of performing edge computing and can handle the computation of neural network kernel functions. This makes it possible to implement a complete end-to-end solution incorporating signal acquisition, digital signal processing, and machine learning for the classification of cardiac arrhythmias on a small wearable device. In this work, we describe the design and implementation of several classifiers for atrial fibrillation detection on a general-purpose ARM Cortex-M4 microcontroller. We used the CMSIS-DSP library, which supports Naïve Bayes and Support Vector Machine classifiers, with different kernel functions. We also developed Python scripts to automatically transfer the Python model (trained in Scikit-learn) to the C environment. To train and evaluate the models, we used part of the data from the PhysioNet/Computing in Cardiology Challenge 2020 and performed simple classification of atrial fibrillation based on heart-rate irregularity. The performance of the classifiers was tested on a general-purpose ARM Cortex-M4 microcontroller (STM32WB55RG). Our study reveals that among the tested classifiers, the SVM classifier with RBF kernel function achieves the highest accuracy of 96.9%, sensitivity of 98.4%, and specificity of 95.8%. The execution time of this classifier was 720 μs per recording. We also discuss the advantages of moving computing tasks to edge devices, including increased power efficiency of the system, improved patient data privacy and security, and reduced overall system operation costs. In addition, we highlight a problem with false-positive detection and unclear significance of device-detected atrial fibrillation.
Davies HJ, Mandic DP, 2023, Rapid extraction of respiratory waveforms from photoplethysmography: A deep corr-encoder approach, Biomedical Signal Processing and Control, Vol: 85, ISSN: 1746-8094
Much of the information related to breathing is contained within the photoplethysmography (PPG) signal, through changes in venous blood flow, heart rate and stroke volume. We aim to leverage this fact, by employing a novel deep learning framework which is a based on a repurposed convolutional autoencoder. Our corr-encoder model aims to encode all of the relevant respiratory information contained within photoplethysmography waveform, and decode it into a waveform that is similar to a gold standard respiratory reference — the Capnogram. The model is employed on two photoplethysmography data sets, namely Capnobase and BIDMC. We show that the model is capable of producing respiratory waveforms that approach the gold standard, while in turn producing state of the art respiratory rate estimates. We also show that when it comes to capturing more advanced respiratory waveform characteristics such as duty cycle, our model is for the most part unsuccessful. A suggested reason for this, in light of a previous study on in-ear PPG, is that the respiratory variations in finger-PPG are far weaker compared with other recording locations. Importantly, our model can perform these waveform estimates in a fraction of a millisecond, giving it the capacity to produce over 6 hours of respiratory waveforms in a single second. Moreover, we attempt to interpret the behaviour of the kernel weights within the model, showing that in part our model intuitively selects different breathing frequencies. The model proposed in this work could help to improve the usefulness of consumer PPG-based wearables for medical applications, where detailed respiratory information is required.
Xu YL, Konstantinidis K, Mandic DP, 2023, Graph-Regularized Tensor Regression: A Domain-Aware Framework for Interpretable Modeling of Multiway Data on Graphs., Neural Comput, Pages: 1-26
Modern data analytics applications are increasingly characterized by exceedingly large and multidimensional data sources. This represents a challenge for traditional machine learning models, as the number of model parameters needed to process such data grows exponentially with the data dimensions, an effect known as the curse of dimensionality. Recently, tensor decomposition (TD) techniques have shown promising results in reducing the computational costs associated with large-dimensional models while achieving comparable performance. However, such tensor models are often unable to incorporate the underlying domain knowledge when compressing high-dimensional models. To this end, we introduce a novel graph-regularized tensor regression (GRTR) framework, whereby domain knowledge about intramodal relations is incorporated into the model in the form of a graph Laplacian matrix. This is then used as a regularization tool to promote a physically meaningful structure within the model parameters. By virtue of tensor algebra, the proposed framework is shown to be fully interpretable, both coefficient-wise and dimension-wise. The GRTR model is validated in a multiway regression setting and compared against competing models and is shown to achieve improved performance at reduced computational costs. Detailed visualizations are provided to help readers gain an intuitive understanding of the employed tensor operations.
Davies HJ, Williams I, Hammour G, et al., 2023, In-ear SpO2 for classification of cognitive workload, IEEE Transactions on Cognitive and Developmental Systems, Vol: 15, Pages: 950-958, ISSN: 2379-8920
The brain is the most metabolically active organ in the body, which increases its metabolic activity, and thus oxygen consumption, with increasing cognitive demand. This motivates us to question whether increased cognitive workload may be measurable through changes in blood oxygen saturation. To this end, we explore the feasibility of cognitive workload tracking based on in-ear SpO2 measurements, which are known to be both robust and exhibit minimal delay. We consider cognitive workload assessment based on an N-back task with randomised order. It is shown that the 2-back and 3-back tasks (high cognitive workload) yield either the lowest median absolute SpO2 or largest median decrease in SpO2 in all of the subjects, indicating a measurable and statistically significant decrease in blood oxygen in response to increased cognitive workload. This makes it possible to classify the four N-back task categories, over 5 second epochs, with a mean accuracy of 90.6%, using features derived from in-ear pulse oximetry, including SpO2, pulse rate and respiration rate. These findings suggest that in-ear SpO2 measurements provide sufficient information for the reliable classification of cognitive workload over short time windows, which promises a new avenue for real time cognitive workload tracking.
Stankovic L, Mandic D, 2023, Convolutional Neural Networks Demystified: A Matched Filtering Perspective-Based Tutorial, IEEE Transactions on Systems, Man, and Cybernetics: Systems, Vol: 53, Pages: 3614-3628, ISSN: 2168-2216
Deep neural networks (DNNs) and especially convolutional neural networks (CNNs) have revolutionized the way we approach the analysis of large quantities of data. However, the largely ad hoc fashion of their development, albeit one reason for their rapid success, has also brought to light the intrinsic limitations of CNNs - in particular, those related to their black box nature. In addition, the ability to 'explain' both the way such systems behave and the results they produce is increasingly becoming an imperative in many practical applications. Therefore, it would be particularly useful to establish physically meaningful mechanisms underpinning the operation of CNNs, thus helping to resolve the issue of interpretability of the processing steps and explain their input-output relationship. To this end, we revisit the operation of CNNs from first principles and show that their very backbone - the convolution operation - represents a matched filter which examines the input for the presence of characteristic patterns in data. Our treatment is based on temporal signals, naturally generated by physical sensors, which admit rigorous analysis through systems science. This serves as a vehicle for a unifying account on the overall functionality of CNNs, whereby both the convolution-activation-pooling chain and learning strategies are shown to admit a compact and elegant interpretation under the umbrella of matched filtering. In addition to helping reveal the physical principles underpinning CNNs and providing an intuitive understanding of their operation, the treatment of CNNs from a matched filtering perspective is also shown to offer a platform to support further developments in this area.
Sau A, 2023, Artificial intelligence-enabled electrocardiogram to distinguish atrioventricular re-entrant tachycardia from atrioventricular nodal re-entrant tachycardia, Cardiovascular Digital Health Journal, Vol: 4, Pages: 60-67, ISSN: 2666-6936
BackgroundAccurately determining arrhythmia mechanism from a 12-lead electrocardiogram (ECG) of supraventricular tachycardia can be challenging. We hypothesized a convolutional neural network (CNN) can be trained to classify atrioventricular re-entrant tachycardia (AVRT) vs atrioventricular nodal re-entrant tachycardia (AVNRT) from the 12-lead ECG, when using findings from the invasive electrophysiology (EP) study as the gold standard.MethodsWe trained a CNN on data from 124 patients undergoing EP studies with a final diagnosis of AVRT or AVNRT. A total of 4962 5-second 12-lead ECG segments were used for training. Each case was labeled AVRT or AVNRT based on the findings of the EP study. The model performance was evaluated against a hold-out test set of 31 patients and compared to an existing manual algorithm.ResultsThe model had an accuracy of 77.4% in distinguishing between AVRT and AVNRT. The area under the receiver operating characteristic curve was 0.80. In comparison, the existing manual algorithm achieved an accuracy of 67.7% on the same test set. Saliency mapping demonstrated the network used the expected sections of the ECGs for diagnoses; these were the QRS complexes that may contain retrograde P waves.ConclusionWe describe the first neural network trained to differentiate AVRT from AVNRT. Accurate diagnosis of arrhythmia mechanism from a 12-lead ECG could aid preprocedural counseling, consent, and procedure planning. The current accuracy from our neural network is modest but may be improved with a larger training dataset.
Liu J, Xu D, Lu Y, et al., 2023, Last-iterate convergence analysis of stochastic momentum methods for neural networks, Neurocomputing, Vol: 527, Pages: 27-35, ISSN: 0925-2312
The stochastic momentum method is a commonly used acceleration technique for solving large-scale stochastic optimization problems. Current convergence results of stochastic momentum methods under non-convex stochastic settings mostly discuss convergence in terms of the random output and minimum output, which requires temporal and spatial statistics of historical data. On the other hand, the last-iterate convergence allows us to avoid storing or selecting past output iterates after each iteration, while maintaining rigour in convergence analysis. To this end, we address the convergence of the last iterate output (called last-iterate convergence) of the stochastic momentum methods for non-convex stochastic optimization problems, in a way which is conformal with traditional optimization theory. For generality, we prove the last-iterate convergence of the stochastic momentum methods under a unified framework, covering both stochastic heavy ball momentum and stochastic Nesterov accelerated gradient momentum, whose momentum factors can be either constant or time-varying coefficients. Finally, the last-iterate convergence of the stochastic momentum methods is verified on the benchmark MNIST and CIFAR-10 datasets. The implementation of SUM is available at: https://github.com/xudp100/SUMhttps://github.com/xudp100/SUM.
Hammour G, Mandic DP, 2023, An In-Ear PPG-Based Blood Glucose Monitor: A Proof-of-Concept Study., Sensors (Basel), Vol: 23
Monitoring diabetes saves lives. To this end, we introduce a novel, unobtrusive, and readily deployable in-ear device for the continuous and non-invasive measurement of blood glucose levels (BGLs). The device is equipped with a low-cost commercially available pulse oximeter whose infrared wavelength (880 nm) is used for the acquisition of photoplethysmography (PPG). For rigor, we considered a full range of diabetic conditions (non-diabetic, pre-diabetic, type I diabetic, and type II diabetic). Recordings spanned nine different days, starting in the morning while fasting, up to a minimum of a two-hour period after eating a carbohydrate-rich breakfast. The BGLs from PPG were estimated using a suite of regression-based machine learning models, which were trained on characteristic features of PPG cycles pertaining to high and low BGLs. The analysis shows that, as desired, an average of 82% of the BGLs estimated from PPG lie in region A of the Clarke error grid (CEG) plot, with 100% of the estimated BGLs in the clinically acceptable CEG regions A and B. These results demonstrate the potential of the ear canal as a site for non-invasive blood glucose monitoring.
Stankovica L, Mandic DP, 2023, Understanding the Basis of Graph Convolutional Neural Networks via an Intuitive Matched Filtering Approach [Lecture Notes], IEEE Signal Processing Magazine, Vol: 40, Pages: 155-165, ISSN: 1053-5888
- Cite
- Citations: 1
Li S, Yu Z, Xiang M, et al., 2023, Reciprocal GAN Through Characteristic Functions (RCF-GAN), IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, Vol: 45, Pages: 2246-2263, ISSN: 0162-8828
Qing Z, Ni J, Li Z, et al., 2023, Performance analysis of the augmented complex-valued least mean kurtosis algorithm, SIGNAL PROCESSING, Vol: 203, ISSN: 0165-1684
Liang Y, Xu D, Zhang N, et al., 2023, Almost sure convergence of stochastic composite objective mirror descent for non-convex non-smooth optimization, OPTIMIZATION LETTERS, ISSN: 1862-4472
Yarici MC, Thornton M, Mandic DP, 2023, Ear-EEG sensitivity modeling for neural sources and ocular artifacts, FRONTIERS IN NEUROSCIENCE, Vol: 16
Talebi SP, Godsill SJ, Mandic DP, 2023, Filtering Structures or alpha-Stable Systems, IEEE CONTROL SYSTEMS LETTERS, Vol: 7, Pages: 553-558, ISSN: 2475-1456
Scalzo B, Stankovic L, Dakovic M, et al., 2023, A class of doubly stochastic shift operators for random graph signals and their boundedness, NEURAL NETWORKS, Vol: 158, Pages: 83-88, ISSN: 0893-6080
Menguc EC, Acr N, Mandic DP, 2023, A Class of Online Censoring Based Quaternion-Valued Least Mean Square Algorithms, IEEE Signal Processing Letters, Vol: 30, Pages: 244-248, ISSN: 1070-9908
Streaming Big Data applications require the means to efficiently utilize large-scale data in an online manner. This issue becomes even more pressing when data are also multidimensional, as is the case with quaternion data streams. To this end, we first introduce the online censoring (OC) based quaternion least mean square (OC-QLMS) and OC-augmented QLMS (OC-AQLMS) algorithms, which censor less informative data in order to reduce computational complexity without severely affecting performance. Next, to censor both the outlier and noninformative data, we also propose the robust OC-QLMS (ROC-QLMS) and ROC-AQLMS. Fixed and adaptive threshold rules are introduced into the proposed OC algorithms to efficiently implement the desired censoring probability in the quaternion domain. The fundamental convergence analysis on the step size for all the proposed algorithms is also presented and the superior properties of the proposed algorithms are demonstrated in system identification scenarios.
Liu C, Ma X, Zhan Y, et al., 2023, Comprehensive Graph Gradual Pruning for Sparse Training in Graph Neural Networks, IEEE Transactions on Neural Networks and Learning Systems, ISSN: 2162-237X
Graph neural networks (GNNs) tend to suffer from high computation costs due to the exponentially increasing scale of graph data and a large number of model parameters, which restricts their utility in practical applications. To this end, some recent works focus on sparsifying GNNs (including graph structures and model parameters) with the lottery ticket hypothesis (LTH) to reduce inference costs while maintaining performance levels. However, the LTH-based methods suffer from two major drawbacks: 1) they require exhaustive and iterative training of dense models, resulting in an extremely large training computation cost, and 2) they only trim graph structures and model parameters but ignore the node feature dimension, where vast redundancy exists. To overcome the above limitations, we propose a comprehensive graph gradual pruning framework termed CGP. This is achieved by designing a during-training graph pruning paradigm to dynamically prune GNNs within one training process. Unlike LTH-based methods, the proposed CGP approach requires no retraining, which significantly reduces the computation costs. Furthermore, we design a cosparsifying strategy to comprehensively trim all the three core elements of GNNs: graph structures, node features, and model parameters. Next, to refine the pruning operation, we introduce a regrowth process into our CGP framework, to reestablish the pruned but important connections. The proposed CGP is evaluated over a node classification task across six GNN architectures, including shallow models graph convolutional network (GCN) and graph attention network (GAT), shallow-but-deep-propagation models simple graph convolution (SGC) and approximate personalized propagation of neural predictions (APPNP), and deep models GCN via initial residual and identity mapping (GCNII) and residual GCN (ResGCN), on a total of 14 real-world graph datasets, including large-scale graph datasets from the challenging Open Graph Benchmark (OGB). Experiments reveal th
Konstantinidis T, Golan M, Xu YL, et al., 2023, Text Mining for Sentiment Analysis in Bond Portfolio Construction
The Efficient Market Hypothesis stipulates that financial instruments within a market perfectly reflect all public information. As the predominant method by which information is disseminated, it is considered that information contained within news influences the value of financial instruments. Thus, if the Efficient Market Hypothesis holds, then understanding the implications of such information is vital in informing portfolio construction. Sentiment analysis is the attempt of a machine to quantify the opinion of natural language. Therefore, this paper visits the Efficient Market Hypothesis by using a novel approach of analysing the relationship between financial news sentiment and corporate bonds in the time frame between 2015 and 2021. Several methods of sentiment analysis are employed to compare the efficacy of traditional lexicon-based knowledge extraction methods and the novel Transformer model approach. Furthermore, the short- and long-term impact of news is evaluated by comparing lagging and decaying signals. The outputs yielded from the sentiment analysis are used as a parameter in portfolio construction ant their returns are examined in light of Sharpe's Ratio to find success in demonstrating the continued viability of the Efficient Market Hypothesis, and the strong relationship between this theory and modern methods of financial sentiment analysis. Ultimately, this paper yields literature leading results with a Sharpe Ratio of 2.1.
Konstantinidis T, Farres AB, Xu YL, et al., 2023, Financial News Classification Model for NLP-based Bond Portfolio Construction
The objective of this paper is to build an accurate classification model for news articles, and to study the effect of different news categories in corporate bond returns. The first step is to show how to classify news articles for traded companies into different defined categories. A classification model using Bag of Words (BoW) and Term Frequency Inverse Document Frequency (TF-IDF) methodology is developed to classify the news into 10 defined categories, and gives an average accuracy of 71%. After web scrapping news articles from 3 different sources to build a dataset of over 550 thousand articles, these are classified into 10 different categories. Natural Language Processing is one of the main techniques used in systematic portfolio construction. Also in the paper, the constructed portfolio based on the sentiment of news articles, by buying the corporate bonds of companies that have positive sentiment in the news, and shorting the bonds of companies with negative sentiment in the news is presented. The effect of each news category in corporate bond returns is studied through these portfolios.
Konstantinidis T, Xu YL, Constantinides TG, et al., 2023, A comparative study on ML-based approaches for Main Entity Detection in Financial Reports
Modern AI technologies which exploit the classification and/or prediction capacities of Deep Neural Architectures demonstrate superior performance to traditional approaches in most cases. However, they come with the unavoidable shortcoming of lack of transparency in their outcomes. This attribute renders them unsuitable for big industrial sectors, such as finance, investment management, etc. Specifically, their "black-box"nature makes them unattractive in cases where human understanding in the decision making process is required and may be legally mandatory. In such cases, traditional (i.e., non-deep learning) ML approaches are still preferred, to minimize for example the presence of false positives. In this context, this paper introduces an unsupervised, trustful, bottom-up probabilistic approach for Named Entity Recognition (NER) in financial reports, while in parallel it provides a comparative study on well-known ML-approaches in terms of their performance. The proposed approach builds on the probability of appearance of representative tokens within the given reports and utilizes Kronecker's Delta and the Total Probability Theorem to construct a probabilistic model that estimates the overall classification probability of a document.
Ibunu S, Moore K, Took CC, et al., 2023, Weight sharing for single-channel LMS, Pages: 185-189
Constraining a group of taps of an adaptive filter to a single value may seem like a futile task, as weight sharing reduces the degree of freedom of the algorithm, and there are no obvious advantages for implementing such an update scheme. On the other hand, weight sharing is popular in deep learning and underpins the success of convolutional neural networks (CNNs) in numerous applications. To this end, we investigate the advantages of weight sharing in single-channel least mean square (LMS), and propose weight sharing LMS (WSLMS) and partial weight sharing LMS (PWS). In particular, we illustrate how weight sharing can lead to numerous benefits such as an enhanced robustness to noise and a computational cost that is independent of the filter length. Simulations support the analysis.
Lo Giudice M, Mammone N, Ieracitano C, et al., 2023, A Convolutional Neural Network Approach for the Classification of Subjects with Epileptic Seizures Versus Psychogenic Non-epileptic Seizures and Control, Based on Automatic Feature Extraction from Empirical Mode Decomposition of Interictal EEG Recordings, Smart Innovation, Systems and Technologies, Pages: 207-214
A reliable data-driven pipeline based on deep learning is introduced to differentiate between individuals with epileptic seizures (ES), psychogenic non-epileptic seizures (PNES), and control subjects (CS) using non-invasive, low-density interictal scalp EEG recordings. The study recruited 42 subjects with ES (new onset), 42 subjects with PNES diagnosed via video-EEG, and 19 CS with normal EEG. Subjects taking psychotropic drugs were excluded to avoid alterations in the EEG signal. The proposed methodology involves automatically extracting features from the 19-channel EEG channels using Empirical Mode Decomposition (EMD) and a customized Convolutional Neural Network (CNN) with a convolutional processing module, rectified linear units (ReLu), and pooling layer to extract and learn relevant features and perform the necessary classification. The CNN displayed excellent classification performance, achieving an accuracy of 85.7%, thereby fostering the use of deep processing systems to aid physicians in challenging clinical situations.
Poushpas S, Normahani P, Kisil I, et al., 2023, Tensor decomposition and machine learning for the detection of arteriovenous fistula stenosis: An initial evaluation., PLoS One, Vol: 18
Duplex ultrasound (DUS) is the most widely used method for surveillance of arteriovenous fistulae (AVF) created for dialysis. However, DUS is poor at predicting AVF outcomes and there is a need for novel methods that can more accurately evaluate multidirectional AVF flow. In this study we aimed to evaluate the feasibility of detecting AVF stenosis using a novel method combining tensor-decomposition of B-mode ultrasound cine loops (videos) of blood flow and machine learning classification. Classification of stenosis was based on the DUS assessment of blood flow volume, vessel diameter size, flow velocity, and spectral waveform features. Real-time B-mode cine loops of the arterial inflow, anastomosis, and venous outflow of the AVFs were analysed. Tensor decompositions were computed from both the 'full-frame' (whole-image) videos and 'cropped' videos (to include areas of blood flow only). The resulting output were labelled for the presence of stenosis, as per the DUS findings, and used as a set of features for classification using a Long Short-Term Memory (LSTM) neural network. A total of 61 out of 66 available videos were used for analysis. The whole-image classifier failed to beat random guessing, achieving a mean area under the receiver operating characteristics (AUROC) value of 0.49 (CI 0.48 to 0.50). In contrast, the 'cropped' video classifier performed better with a mean AUROC of 0.82 (CI 0.66 to 0.96), showing promising predictive power despite the small size of the dataset. The combined application of tensor decomposition and machine learning are promising for the detection of AVF stenosis and warrant further investigation.
Lo Giudice M, Ferlazzo E, Mammone N, et al., 2022, Convolutional Neural Network Classification of Rest EEG Signals among People with Epilepsy, Psychogenic Non Epileptic Seizures and Control Subjects, INTERNATIONAL JOURNAL OF ENVIRONMENTAL RESEARCH AND PUBLIC HEALTH, Vol: 19
Roa-Vicens J, Xu YL, Silva R, et al., 2022, Graph and tensor-train recurrent neural networks for high-dimensional models of limit order books, Pages: 207-213
Recurrent neural networks (RNNs) have proven to be particularly effective for the paradigms of learning and modelling time series. However, sequential data of high dimensions are considerably more difficult and computationally expensive to model, as the number of parameters required to train the RNN grows exponentially with data dimensionality. This is also the case with time series from limit order books, the electronic registries where prices of securities are formed in public markets. To this end, tensorization of neural networks provides an efficient method to reduce the number of model parameters, and has been applied successfully to high-dimensional series such as video sequences and financial time series, for example, using tensor-train RNNs (TTRNNs). However, such TTRNNs suffer from a number of shortcomings, including: (i) model sensitivity to the ordering of core tensor contractions; (ii) training sensitivity to weight initialization; and (iii) exploding or vanishing gradient problems due to the recurrent propagation through the tensor-train topology. Recent studies showed that embedding a multi-linear graph filter to model RNN states (Recurrent Graph Tensor Network, RGTN) provides enhanced flexibility and expressive power to tensor networks, while mitigating the shortcomings of TTRNNs. In this paper, we demonstrate the advantages arising from the use of graph filters to model limit order book sequences of high dimension as compared with the state-of-the-art benchmarks. It is shown that the combination of the graph module (to mitigate problematic gradients) with the radial structure (to make the tensor network architecture flexible) results in substantial improvements in output variance, training time and number of parameters required, without any sacrifice in accuracy.
Powezka K, Pettipher A, Hemakom A, et al., 2022, A Pilot Study of Heart Rate Variability Synchrony as a Marker of Intraoperative Surgical Teamwork and Its Correlation to the Length of Procedure, SENSORS, Vol: 22
Kisil I, Calvi GG, Konstantinidis K, et al., 2022, Accelerating Tensor Contraction Products via Tensor-Train Decomposition [Tips & Tricks], IEEE SIGNAL PROCESSING MAGAZINE, Vol: 39, Pages: 63-70, ISSN: 1053-5888
Xiao H, Chanwimalueang T, Mandic DP, 2022, Multivariate Multiscale Cosine Similarity Entropy and Its Application to Examine Circularity Properties in Division Algebras, ENTROPY, Vol: 24
- Author Web Link
- Cite
- Citations: 1
Sau A, Ibrahim S, Ahmed A, et al., 2022, Artificial intelligence-enabled electrocardiogram to distinguish cavotricuspid isthmus dependence from other atrial tachycardia mechanisms, European Heart Journal – Digital Health, Vol: 3, Pages: 405-414, ISSN: 2634-3916
Aims:Accurately determining atrial arrhythmia mechanisms from a 12-lead electrocardiogram (ECG) can be challenging. Given the high success rate of cavotricuspid isthmus (CTI) ablation, identification of CTI-dependent typical atrial flutter (AFL) is important for treatment decisions and procedure planning. We sought to train a convolutional neural network (CNN) to classify CTI-dependent AFL vs. non-CTI dependent atrial tachycardia (AT), using data from the invasive electrophysiology (EP) study as the gold standard.Methods and results:We trained a CNN on data from 231 patients undergoing EP studies for atrial tachyarrhythmia. A total of 13 500 five-second 12-lead ECG segments were used for training. Each case was labelled CTI-dependent AFL or non-CTI-dependent AT based on the findings of the EP study. The model performance was evaluated against a test set of 57 patients. A survey of electrophysiologists in Europe was undertaken on the same 57 ECGs. The model had an accuracy of 86% (95% CI 0.77–0.95) compared to median expert electrophysiologist accuracy of 79% (range 70–84%). In the two thirds of test set cases (38/57) where both the model and electrophysiologist consensus were in agreement, the prediction accuracy was 100%. Saliency mapping demonstrated atrial activation was the most important segment of the ECG for determining model output.Conclusion:We describe the first CNN trained to differentiate CTI-dependent AFL from other AT using the ECG. Our model matched and complemented expert electrophysiologist performance. Automated artificial intelligence-enhanced ECG analysis could help guide treatment decisions and plan ablation procedures for patients with organized atrial arrhythmias.
Thornton M, Mandic D, Reichenbach T, 2022, Robust decoding of the speech envelope from EEG recordings through deep neural networks, JOURNAL OF NEURAL ENGINEERING, Vol: 19, ISSN: 1741-2560
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.