Deep learning and the electrocardiogram: review of the current state-of-the-art

Rhythm-Ready: Harnessing Smart Devices to Detect and Manage Arrhythmias

PURPOSE OF REVIEW

To survey recent progress in the application of implantable and wearable sensors to detection and management of cardiac arrhythmias.

RECENT FINDINGS

Sensor-enabled strategies are critical for the detection, prediction and management of arrhythmias. In the last several years, great innovation has occurred in the types of devices (implanted and wearable) that are available and the data they collect. The integration of artificial intelligence solutions into sensor-enabled strategies has set the stage for a new generation of smart devices that augment the human clinician. Smart devices enhanced by new sensor technologies and Artificial Intelligence (AI) algorithms promise to reshape the care of cardiac arrhythmias.

Deep learning assists early-detection of hypertension-mediated heart change on ECG signals

Arterial hypertension is a major risk factor for cardiovascular diseases. While cardiac ultrasound is a typical way to diagnose hypertension-mediated heart change, it often fails to detect early subtle structural changes. Electrocardiogram(ECG) represents electrical activity of heart muscle, affected by the changes in heart's structure. It is crucial to explore whether ECG can capture slight signals of hypertension-mediated heart change. However, reading ECG records is complex and some signals are too subtle to be captured by cardiologist's visual inspection. In this study, we designed a deep learning model to predict hypertension on ECG signals and then to identify hypertension-associated ECG segments. From The First Affiliated Hospital of Xiamen University, we collected 210,120 10-s 12-lead ECGs using the FX-8322 manufactured by FUKUDA and 812 ECGs using the RAGE-12 manufactured by NALONG. We proposed a deep learning framework, including MML-Net, a multi-branch, multi-scale LSTM neural network to evaluate the potential of ECG signals to detect hypertension, and ECG-XAI, an ECG-oriented wave-alignment AI explanation pipeline to identify hypertension-associated ECG segments. MML-Net achieved an 82% recall and an 87% precision in the testing, and an 80% recall and an 82% precision in the independent testing. In contrast, experienced clinical cardiologists typically attain recall rates ranging from 30 to 50% by visual inspection. The experiments demonstrate that ECG signals are sensitive to slight changes in heart structure caused by hypertension. ECG-XAI detects that R-wave and P-wave are the hypertension-associated ECG segments. The proposed framework has the potential to facilitate early diagnosis of heart change.

Enhancing origin prediction: deep learning model for diagnosing premature ventricular contractions with dual-rhythm analysis focused on cardiac rotation

AIMS

Several algorithms can differentiate inferior axis premature ventricular contractions (PVCs) originating from the right side and left side on 12-lead electrocardiograms (ECGs). However, it is unclear whether distinguishing the origin should rely solely on PVC or incorporate sinus rhythm (SR). We compared the dual-rhythm model (incorporating both SR and PVC) to the PVC model (using PVC alone) and quantified the contribution of each ECG lead in predicting the PVC origin for each cardiac rotation.

METHODS AND RESULTS

This multicentre study enrolled 593 patients from 11 centres-493 from Japan and Germany, and 100 from Belgium, which were used as the external validation data set. Using a hybrid approach combining a Resnet50-based convolutional neural network and a transformer model, we developed two variants-the PVC and dual-rhythm models-to predict PVC origin. In the external validation data set, the dual-rhythm model outperformed the PVC model in accuracy (0.84 vs. 0.74, respectively; P < 0.01), precision (0.73 vs. 0.55, respectively; P < 0.01), specificity (0.87 vs. 0.68, respectively; P < 0.01), area under the receiver operating characteristic curve (0.91 vs. 0.86, respectively; P = 0.03), and F1-score (0.77 vs. 0.68, respectively; P = 0.03). The contributions to PVC origin prediction were 77.3% for PVC and 22.7% for the SR. However, in patients with counterclockwise rotation, SR had a greater contribution in predicting the origin of right-sided PVC.

CONCLUSION

Our deep learning-based model, incorporating both PVC and SR morphologies, resulted in a higher prediction accuracy for PVC origin, considering SR is particularly important for predicting right-sided origin in patients with counterclockwise rotation.

Machine Learning Algorithm to Predict Atrial Fibrillation Using Serial 12-Lead ECGs Based on Left Atrial Remodeling

BACKGROUND

We hypothesized that analysis of serial ECGs could predict new-onset atrial fibrillation (AF) more accurately than analysis of a single ECG by detecting the subtle cardiac remodeling that occurs immediately before AF occurrence. Our aim in this study was to compare the performance of 2 types of machine learning (ML) algorithms.

METHODS AND RESULTS

Standard 12-lead ECGs of patients selected by cardiologists between January 2010 and May 2021 were used for ML model development. Two ML models (single ECG and serial ECG) were developed using a light gradient boosting machine-learning algorithm. Model performance was evaluated based on the area under the receiver operating characteristic curve, sensitivity, specificity, accuracy, and F1 score. We trained the ML models on 415 964 ECGs from 176 090 patients. When testing the 2 ML models using external validation data sets, the performance of the serial-ML model was significantly better than that of the single-ML model for predicting new-onset AF (single- versus serial-ML model: sensitivity 0.744 versus 0.810; specificity 0.742 versus 0.822; accuracy 0.743 versus 0.816; F1 score 0.743 versus 0.815; area under the receiver operating characteristic curve 0.812 versus 0.880; P<0.001). The Shapley Additive Explanations analysis ranked P-wave duration and amplitude among the top 10 ECG parameters.

CONCLUSIONS

An ML model based on serial ECGs from an individual had greater ability to predict new-onset AF than the ML model based on a single ECG. P-wave morphologies were associated with future AF prediction.

Novel interpretable Feature set extraction and classification for accurate atrial fibrillation detection from ECGs

OBJECTIVE

We present a novel method for detecting atrial fibrillation (AFib) by analyzing Lead II electrocardiograms (ECGs) using a unique set of features.

METHODS

For this purpose, we used specific signal processing techniques, such as proper orthogonal decomposition, continuous wavelet transforms, discrete cosine transform, and standard cross-correlation, to extract 48 features from the ECGs. Thus, our approach aims to more effectively capture AFib signatures, such as beat-to-beat variability and fibrillatory waves, than traditional metrics. Moreover, our features were designed to be physiologically interpretable. Subsequently, we incorporated an XGBoost-based ECG classifier to train and evaluate it using the publicly available 'Training' dataset of the 2017 PhysioNet Challenge, which includes 'Normal,' 'AFib,' 'Other,' and 'Noisy' ECG categories.

RESULTS

Our method achieved an accuracy of 96 % and an F1-score of 0.83 for 'AFib' detection and 80 % accuracy and 0.85 F1-score for 'Normal' ECGs. Finally, we compared our method's performance with the top-classifiers from the 2017 PhysioNet Challenge, namely ENCASE, Random Forest, and Cascaded Binary. As reported by Clifford et al., 2017, these three best performing models scored 0.80, 0.83, 0.82, in terms of F1-score for 'AFib' detection, respectively.

CONCLUSION

Despite using significantly fewer features than the competition's state-of-the-art ECG detection algorithms (48 vs. 150-622), our model achieved a comparable F1-score of 0.83 for successful 'AFib' detection.

SIGNIFICANCE

The interpretable features specifically designed for 'AFib' detection results in our method's adaptability in clinical settings for real-time arrhythmia detection and is even effective with short ECGs (<10 heartbeats).

Detection and Classification of electrocardiography using hybrid deep learning models

BACKGROUND

Electrocardiography (ECGs) has been a vital tool for cardiovascular disease (CVD) diagnosis, which visually depicts the heart's electrical activity. To enhance automatic classification between normal and diseased ECG, it is essential to extract consistent and qualitative features.

METHODS

Precision of ECG classification through hybrid Deep Learning (DL) approach leverages both Convolutional Neural Network (CNN) architecture and Variational Autoencoder (VAE) techniques. By combining these methods, we aim to achieve more accurate and robust ECG interpretation. The method is trained and tested over PTB-XL dataset, which contains 21,799 with 12-lead ECGs from 18,869 patients, each spanning 10 seconds. The classification evaluation of 5 super-classes and 23 sub-classes of CVD, with the proposed CNN-VAE model is compared.

RESULTS

The classification of various CVD had resulted with the highest accuracy of 98.51%, specificity of 98.12%, sensitivity 97.9% and F1-score 97.95%. We have also achieved the minimum false positive and false negative rates as 2.07 and 1.87 respectively during validation. The results are validated upon the annotations given by individual cardiologists, who assigned potentially multiple ECG statements to each record.

CONCLUSION

When compared to other deep learning methods, our suggested CNN-VAE model performs significantly better in testing phase. This study proposes a new architecture of combining CNN-VAE for CVD classification from ECG data, this can help the clinicians to identify the disease earlier and carry further treatment. The CNN-VAE model can better characterize input signals due to its hybrid architecture.

An artificial intelligence-enabled electrocardiogram algorithm for the prediction of left atrial low-voltage areas in persistent atrial fibrillation

OBJECTIVES

We aimed to construct an artificial intelligence-enabled electrocardiogram (ECG) algorithm that can accurately predict the presence of left atrial low-voltage areas (LVAs) in patients with persistent atrial fibrillation.

METHODS

The study included 587 patients with persistent atrial fibrillation who underwent catheter ablation procedures between March 2012 and December 2023 and 942 scanned images of 12-lead ECGs obtained before the ablation procedures were performed. Artificial intelligence-based algorithms were used to construct models for predicting the presence of LVAs. The DR-FLASH and APPLE clinical scores for LVA prediction were calculated. We used a receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis to evaluate model performance.

RESULTS

The data obtained from the participants were split into training (n = 469), validation (n = 58), and test sets (n = 60). LVAs were detected in 53.7% of all participants. Using ECG alone, the deep learning algorithm achieved an area under the ROC curve (AUROC) of 0.752, outperforming both the DR-FLASH score (AUROC = 0.610) and the APPLE score (AUROC = 0.510). The random forest classification model, which integrated a probabilistic deep learning model and clinical features, showed a maximum AUROC of 0.759. Moreover, the ECG-based deep learning algorithm for predicting extensive LVAs achieved an AUROC of 0.775, with a sensitivity of 0.816 and a specificity of 0.896. The random forest classification model for predicting extensive LVAs achieved an AUROC of 0.897, with a sensitivity of 0.862, and a specificity of 0.935.

CONCLUSION

The deep learning model based exclusively on ECG data and the machine learning model that combined a probabilistic deep learning model and clinical features both predicted the presence of LVAs with a higher degree of accuracy than the DR-FLASH and the APPLE risk scores.

Deep Neural Networks and Tabular Data: A Survey

Heterogeneous tabular data are the most commonly used form of data and are essential for numerous critical and computationally demanding applications. On homogeneous datasets, deep neural networks have repeatedly shown excellent performance and have therefore been widely adopted. However, their adaptation to tabular data for inference or data generation tasks remains highly challenging. To facilitate further progress in the field, this work provides an overview of state-of-the-art deep learning methods for tabular data. We categorize these methods into three groups: data transformations, specialized architectures, and regularization models. For each of these groups, our work offers a comprehensive overview of the main approaches. Moreover, we discuss deep learning approaches for generating tabular data and also provide an overview over strategies for explaining deep models on tabular data. Thus, our first contribution is to address the main research streams and existing methodologies in the mentioned areas while highlighting relevant challenges and open research questions. Our second contribution is to provide an empirical comparison of traditional machine learning methods with 11 deep learning approaches across five popular real-world tabular datasets of different sizes and with different learning objectives. Our results, which we have made publicly available as competitive benchmarks, indicate that algorithms based on gradient-boosted tree ensembles still mostly outperform deep learning models on supervised learning tasks, suggesting that the research progress on competitive deep learning models for tabular data is stagnating. To the best of our knowledge, this is the first in-depth overview of deep learning approaches for tabular data; as such, this work can serve as a valuable starting point to guide researchers and practitioners interested in deep learning with tabular data.

Biocompatible Electronic Skins for Cardiovascular Health Monitoring

Cardiovascular diseases represent a significant threat to the overall well-being of the global population. Continuous monitoring of vital signs related to cardiovascular health is essential for improving daily health management. Currently, there has been remarkable proliferation of technology focused on collecting data related to cardiovascular diseases through daily electronic skin monitoring. However, concerns have arisen regarding potential skin irritation and inflammation due to the necessity for prolonged wear of wearable devices. To ensure comfortable and uninterrupted cardiovascular health monitoring, the concept of biocompatible electronic skin has gained substantial attention. In this review, biocompatible electronic skins for cardiovascular health monitoring are comprehensively summarized and discussed. The recent achievements of biocompatible electronic skin in cardiovascular health monitoring are introduced. Their working principles, fabrication processes, and performances in sensing technologies, materials, and integration systems are highlighted, and comparisons are made with other electronic skins used for cardiovascular monitoring. In addition, the significance of integrating sensing systems and the updating wireless communication for the development of the smart medical field is explored. Finally, the opportunities and challenges for wearable electronic skin are also examined.

Artificial Intelligence in Sports Medicine: Reshaping Electrocardiogram Analysis for Athlete Safety-A Narrative Review

Artificial Intelligence (AI) is redefining electrocardiogram (ECG) analysis in pre-participation examination (PPE) of athletes, enhancing the detection and monitoring of cardiovascular health. Cardiovascular concerns, including sudden cardiac death, pose significant risks during sports activities. Traditional ECG, essential yet limited, often fails to distinguish between benign cardiac adaptations and serious conditions. This narrative review investigates the application of machine learning (ML) and deep learning (DL) in ECG interpretation, aiming to improve the detection of arrhythmias, channelopathies, and hypertrophic cardiomyopathies. A literature review over the past decade, sourcing from PubMed and Google Scholar, highlights the growing adoption of AI in sports medicine for its precision and predictive capabilities. AI algorithms excel at identifying complex cardiac patterns, potentially overlooked by traditional methods, and are increasingly integrated into wearable technologies for continuous monitoring. Overall, by offering a comprehensive overview of current innovations and outlining future advancements, this review supports sports medicine professionals in merging traditional screening methods with state-of-the-art AI technologies. This approach aims to enhance diagnostic accuracy and efficiency in athlete care, promoting early detection and more effective monitoring through AI-enhanced ECG analysis within athlete PPEs.

Myocardial infarction detection method based on the continuous T-wave area feature and multi-lead-fusion deep features

Objective.Myocardial infarction (MI) is one of the most threatening cardiovascular diseases. This paper aims to explore a method for using an algorithm to autonomously classify MI based on the electrocardiogram (ECG).Approach.A detection method of MI that fuses continuous T-wave area (C_TWA) feature and ECG deep features is proposed. This method consists of three main parts: (1) The onset of MI is often accompanied by changes in the shape of the T-wave in the ECG, thus the area of the T-wave displayed on different heartbeats will be quite different. The adaptive sliding window method is used to detect the start and end of the T-wave, and calculate the C_TWA on the same ECG record. Additionally, the coefficient of variation of C_TWA is defined as the C_TWA feature of the ECG. (2) The multi lead fusion convolutional neural network was implemented to extract the deep features of the ECG. (3) The C_TWA feature and deep features of the ECG were fused by soft attention, and then inputted into the multi-layer perceptron to obtain the detection result.Main results.According to the inter-patient paradigm, the proposed method reached a 97.67% accuracy, 96.59% precision, and 98.96% recall on the PTB dataset, as well as reached 93.15% accuracy, 93.20% precision, and 95.14% recall on the clinical dataset.Significance.This method accurately extracts the feature of the C_TWA, and combines the deep features of the signal, thereby improving the detection accuracy and achieving favorable results on clinical datasets.

Development and validation of machine learning algorithms based on electrocardiograms for cardiovascular diagnoses at the population level

Artificial intelligence-enabled electrocardiogram (ECG) algorithms are gaining prominence for the early detection of cardiovascular (CV) conditions, including those not traditionally associated with conventional ECG measures or expert interpretation. This study develops and validates such models for simultaneous prediction of 15 different common CV diagnoses at the population level. We conducted a retrospective study that included 1,605,268 ECGs of 244,077 adult patients presenting to 84 emergency departments or hospitals, who underwent at least one 12-lead ECG from February 2007 to April 2020 in Alberta, Canada, and considered 15 CV diagnoses, as identified by International Classification of Diseases, 10th revision (ICD-10) codes: atrial fibrillation (AF), supraventricular tachycardia (SVT), ventricular tachycardia (VT), cardiac arrest (CA), atrioventricular block (AVB), unstable angina (UA), ST-elevation myocardial infarction (STEMI), non-STEMI (NSTEMI), pulmonary embolism (PE), hypertrophic cardiomyopathy (HCM), aortic stenosis (AS), mitral valve prolapse (MVP), mitral valve stenosis (MS), pulmonary hypertension (PHTN), and heart failure (HF). We employed ResNet-based deep learning (DL) using ECG tracings and extreme gradient boosting (XGB) using ECG measurements. When evaluated on the first ECGs per episode of 97,631 holdout patients, the DL models had an area under the receiver operating characteristic curve (AUROC) of <80% for 3 CV conditions (PTE, SVT, UA), 80-90% for 8 CV conditions (CA, NSTEMI, VT, MVP, PHTN, AS, AF, HF) and an AUROC > 90% for 4 diagnoses (AVB, HCM, MS, STEMI). DL models outperformed XGB models with about 5% higher AUROC on average. Overall, ECG-based prediction models demonstrated good-to-excellent prediction performance in diagnosing common CV conditions.

Impact of ECG data format on the performance of machine learning models for the prediction of myocardial infarction

Background We aim to determine which electrocardiogram (ECG) data format is optimal for ML modelling, in the context of myocardial infarction prediction. We will also address the auxiliary objective of evaluating the viability of using digitised ECG signals for ML modelling. Methods Two ECG arrangements displaying 10s and 2.5 s of data for each lead were used. For each arrangement, conservative and speculative data cohorts were generated from the PTB-XL dataset. All ECGs were represented in three different data formats: Signal ECGs, Image ECGs, and Extracted Signal ECGs, with 8358 and 11,621 ECGs in the conservative and speculative cohorts, respectively. ML models were trained using the three data formats in both data cohorts. Results For ECGs that contained 10s of data, Signal and Extracted Signal ECGs were optimal and statistically similar, with AUCs [95% CI] of 0.971 [0.961, 0.981] and 0.974 [0.965, 0.984], respectively, for the conservative cohort; and 0.931 [0.918, 0.945] and 0.919 [0.903, 0.934], respectively, for the speculative cohort. For ECGs that contained 2.5 s of data, the Image ECG format was optimal, with AUCs of 0.960 [0.948, 0.973] and 0.903 [0.886, 0.920], for the conservative and speculative cohorts, respectively. Conclusion When available, the Signal ECG data should be preferred for ML modelling. If not, the optimal format depends on the data arrangement within the ECG: If the Image ECG contains 10s of data for each lead, the Extracted Signal ECG is optimal, however, if it only uses 2.5 s, then using the Image ECG data is optimal for ML performance.

Super-resolution techniques for biomedical applications and challenges

Super-resolution (SR) techniques have revolutionized the field of biomedical applications by detailing the structures at resolutions beyond the limits of imaging or measuring tools. These techniques have been applied in various biomedical applications, including microscopy, magnetic resonance imaging (MRI), computed tomography (CT), X-ray, electroencephalogram (EEG), ultrasound, etc. SR methods are categorized into two main types: traditional non-learning-based methods and modern learning-based approaches. In both applications, SR methodologies have been effectively utilized on biomedical images, enhancing the visualization of complex biological structures. Additionally, these methods have been employed on biomedical data, leading to improvements in computational precision and efficiency for biomedical simulations. The use of SR techniques has resulted in more detailed and accurate analyses in diagnostics and research, essential for early disease detection and treatment planning. However, challenges such as computational demands, data interpretation complexities, and the lack of unified high-quality data persist. The article emphasizes these issues, underscoring the need for ongoing development in SR technologies to further improve biomedical research and patient care outcomes.

ECG-surv: A deep learning-based model to predict time to 1-year mortality from 12-lead electrocardiogram

BACKGROUND

Electrocardiogram (ECG) abnormalities have demonstrated potential as prognostic indicators of patient survival. However, the traditional statistical approach is constrained by structured data input, limiting its ability to fully leverage the predictive value of ECG data in prognostic modeling.

METHODS

This study aims to introduce and evaluate a deep-learning model to simultaneously handle censored data and unstructured ECG data for survival analysis. We herein introduce a novel deep neural network called ECG-surv, which includes a feature extraction neural network and a time-to-event analysis neural network. The proposed model is specifically designed to predict the time to 1-year mortality by extracting and analyzing unique features from 12-lead ECG data. ECG-surv was evaluated using both an independent test set and an external set, which were collected using different ECG devices.

RESULTS

The performance of ECG-surv surpassed that of the Cox proportional model, which included demographics and ECG waveform parameters, in predicting 1-year all-cause mortality, with a significantly higher concordance index (C-index) in ECG-surv than in the Cox model using both the independent test set (0.860 [95% CI: 0.859- 0.861] vs. 0.796 [95% CI: 0.791- 0.800]) and the external test set (0.813 [95% CI: 0.807- 0.814] vs. 0.764 [95% CI: 0.755- 0.770]). ECG-surv also demonstrated exceptional predictive ability for cardiovascular death (C-index of 0.891 [95% CI: 0.890- 0.893]), outperforming the Framingham risk Cox model (C-index of 0.734 [95% CI: 0.715-0.752]).

CONCLUSION

ECG-surv effectively utilized unstructured ECG data in a survival analysis. It outperformed traditional statistical approaches in predicting 1-year all-cause mortality and cardiovascular death, which makes it a valuable tool for predicting patient survival.

Preclinical identification of acute coronary syndrome without high sensitivity troponin assays using machine learning algorithms

Preclinical management of patients with acute chest pain and their identification as candidates for urgent coronary revascularization without the use of high sensitivity troponin essays remains a critical challenge in emergency medicine. We enrolled 2760 patients (average age 70 years, 58.6% male) with chest pain and suspected ACS, who were admitted to the Emergency Department of the University Hospital Tübingen, Germany, between August 2016 and October 2020. Using 26 features, eight Machine learning models (non-deep learning models) were trained with data from the preclinical rescue protocol and compared to the "TropOut" score (a modified version of the "preHEART" score which consists of history, ECG, age and cardiac risk but without troponin analysis) to predict major adverse cardiac event (MACE) and acute coronary artery occlusion (ACAO). In our study population MACE occurred in 823 (29.8%) patients and ACAO occurred in 480 patients (17.4%). Interestingly, we found that all machine learning models outperformed the "TropOut" score. The VC and the LR models showed the highest area under the receiver operating characteristic (AUROC) for predicting MACE (AUROC = 0.78) and the VC showed the highest AUROC for predicting ACAO (AUROC = 0.81). A SHapley Additive exPlanations (SHAP) analyses based on the XGB model showed that presence of ST-elevations in the electrocardiogram (ECG) were the most important features to predict both endpoints.

Edge AI Model Deployed for Real-Time Detection of Atrial Fibrillation Risk during Sinus Rhythm

Objectives: The study aimed to develop a deep learning-based edge AI model deployed on electrocardiograph (ECG) devices for the real-time detection of atrial fibrillation (AF) risk during sinus rhythm (SR) using standard 10 s, 12-lead electrocardiograms (ECGs). Methods: A novel approach was used to convert standard 12-lead ECGs into binary images for model input, and a lightweight convolutional neural network (CNN)-based model was trained using data collected by the Japan Agency for Medical and Research Development (AMED) between 2019 and 2022. Patients over 40 years old with digital, SR ECGs were retrospectively enrolled and divided into AF and non-AF groups. The data labeling was supervised by cardiologists. The dataset was randomly allocated into training, validation, and internal testing datasets. External testing was conducted on data collected from other hospitals. Results: The best-trained model achieved an AUC of 0.82 and 0.80, sensitivity of 79.5% and 72.3%, specificity of 77.8% and 77.7%, precision of 78.2% and 76.4%, and overall accuracy of 78.6% and 75.0% in the internal and external testing datasets, respectively. The deployed model and app package utilized 2.5 MB and 40 MB of the available ROM and RAM capacity on the edge ECG device, correspondingly. The processing time for AF risk detection was approximately 2 s. Conclusions: The model maintains comparable performance and improves its suitability for deployment on resource-constrained ECG devices, thereby expanding its potential impact to a wide range of healthcare settings. Its successful deployment enables real-time AF risk detection during SR, allowing for timely intervention to prevent AF-related serious consequences like stroke and premature death.

Transforming the cardiometabolic disease landscape: Multimodal AI-powered approaches in prevention and management

The rise of artificial intelligence (AI) has revolutionized various scientific fields, particularly in medicine, where it has enabled the modeling of complex relationships from massive datasets. Initially, AI algorithms focused on improved interpretation of diagnostic studies such as chest X-rays and electrocardiograms in addition to predicting patient outcomes and future disease onset. However, AI has evolved with the introduction of transformer models, allowing analysis of the diverse, multimodal data sources existing in medicine today. Multimodal AI holds great promise in more accurate disease risk assessment and stratification as well as optimizing the key driving factors in cardiometabolic disease: blood pressure, sleep, stress, glucose control, weight, nutrition, and physical activity. In this article we outline the current state of medical AI in cardiometabolic disease, highlighting the potential of multimodal AI to augment personalized prevention and treatment strategies in cardiometabolic disease.

ECG-based data-driven solutions for diagnosis and prognosis of cardiovascular diseases: A systematic review

Cardiovascular diseases (CVD) are a leading cause of death globally, and result in significant morbidity and reduced quality of life. The electrocardiogram (ECG) plays a crucial role in CVD diagnosis, prognosis, and prevention; however, different challenges still remain, such as an increasing unmet demand for skilled cardiologists capable of accurately interpreting ECG. This leads to higher workload and potential diagnostic inaccuracies. Data-driven approaches, such as machine learning (ML) and deep learning (DL) have emerged to improve existing computer-assisted solutions and enhance physicians' ECG interpretation of the complex mechanisms underlying CVD. However, many ML and DL models used to detect ECG-based CVD suffer from a lack of explainability, bias, as well as ethical, legal, and societal implications (ELSI). Despite the critical importance of these Trustworthy Artificial Intelligence (AI) aspects, there is a lack of comprehensive literature reviews that examine the current trends in ECG-based solutions for CVD diagnosis or prognosis that use ML and DL models and address the Trustworthy AI requirements. This review aims to bridge this knowledge gap by providing a systematic review to undertake a holistic analysis across multiple dimensions of these data-driven models such as type of CVD addressed, dataset characteristics, data input modalities, ML and DL algorithms (with a focus on DL), and aspects of Trustworthy AI like explainability, bias and ethical considerations. Additionally, within the analyzed dimensions, various challenges are identified. To these, we provide concrete recommendations, equipping other researchers with valuable insights to understand the current state of the field comprehensively.

An Arrhythmia classification approach via deep learning using single-lead ECG without QRS wave detection

Arrhythmia, a frequently encountered and life-threatening cardiac disorder, can manifest as a transient or isolated event. Traditional automatic arrhythmia detection methods have predominantly relied on QRS-wave signal detection. Contemporary research has focused on the utilization of wearable devices for continuous monitoring of heart rates and rhythms through single-lead electrocardiogram (ECG), which holds the potential to promptly detect arrhythmias. However, in this study, we employed a convolutional neural network (CNN) to classify distinct arrhythmias without QRS wave detection step. The ECG data utilized in this study were sourced from the publicly accessible PhysioNet databases. Taking into account the impact of the duration of ECG signal on accuracy, this study trained one-dimensional CNN models with 5-s and 10-s segments, respectively, and compared their results. In the results, the CNN model exhibited the capability to differentiate between Normal Sinus Rhythm (NSR) and various arrhythmias, including Atrial Fibrillation (AFIB), Atrial Flutter (AFL), Wolff-Parkinson-White syndrome (WPW), Ventricular Fibrillation (VF), Ventricular Tachycardia (VT), Ventricular Flutter (VFL), Mobitz II AV Block (MII), and Sinus Bradycardia (SB). Both 10-s and 5-s ECG segments exhibited comparable results, with an average classification accuracy of 97.31%. It reveals the feasibility of utilizing even shorter 5-s recordings for detecting arrhythmias in everyday scenarios. Detecting arrhythmias with a single lead aligns well with the practicality of wearable devices for daily use, and shorter detection times also align with their clinical utility in emergency situations.

The Danish Nationwide Electrocardiogram (ECG) Cohort

The electrocardiogram (ECG) is a non-invasive diagnostic tool holding significant clinical importance in the diagnosis and risk stratification of cardiac disease. However, access to large-scale, population-based digital ECG data for research purposes remains limited and challenging. Consequently, we established the Danish Nationwide ECG Cohort to provide data from standard 12-lead digital ECGs in both pre- and in-hospital settings, which can be linked to comprehensive Danish nationwide administrative registers on health and social data with long-term follow-up. The Danish Nationwide ECG Cohort is an open real-world cohort including all patients with at least one digital pre- or in-hospital ECG in Denmark from January 01, 2000, to December 31, 2021. The cohort includes data on standardized and uniform ECG diagnostic statements and ECG measurements including global parameters as well as lead-specific measures of waveform amplitudes, durations, and intervals. Currently, the cohort comprises 2,485,987 unique patients with a median age at the first ECG of 57 years (25th-75th percentiles, 40-71 years; males, 48%), resulting in a total of 11,952,430 ECGs. In conclusion, the Danish Nationwide ECG Cohort represents a novel and extensive population-based digital ECG dataset for cardiovascular research, encompassing both pre- and in-hospital settings. The cohort contains ECG diagnostic statements and ECG measurements that can be linked to various nationwide health and social registers without loss to follow-up.

Artificial intelligence-based identification of left ventricular systolic dysfunction from 12-lead electrocardiograms: external validation and advanced application of an existing model

Aims

The diagnostic application of artificial intelligence (AI)-based models to detect cardiovascular diseases from electrocardiograms (ECGs) evolves, and promising results were reported. However, external validation is not available for all published algorithms. The aim of this study was to validate an existing algorithm for the detection of left ventricular systolic dysfunction (LVSD) from 12-lead ECGs.

Methods and results

Patients with digitalized data pairs of 12-lead ECGs and echocardiography (at intervals of ≤7 days) were retrospectively selected from the Heart Center Leipzig ECG and electronic medical records databases. A previously developed AI-based model was applied to ECGs and calculated probabilities for LVSD. The area under the receiver operating characteristic curve (AUROC) was computed overall and in cohorts stratified for baseline and ECG characteristics. Repeated echocardiography studies recorded ≥3 months after index diagnostics were used for follow-up (FU) analysis. At baseline, 42 291 ECG-echocardiography pairs were analysed, and AUROC for LVSD detection was 0.88. Sensitivity and specificity were 82% and 77% for the optimal LVSD probability cut-off based on Youden's J. AUROCs were lower in ECG subgroups with tachycardia, atrial fibrillation, and wide QRS complexes. In patients without LVSD at baseline and available FU, model-generated high probability for LVSD was associated with a four-fold increased risk of developing LVSD during FU.

Conclusion

We provide the external validation of an existing AI-based ECG-analysing model for the detection of LVSD with robust performance metrics. The association of false positive LVSD screenings at baseline with a deterioration of ventricular function during FU deserves a further evaluation in prospective trials.

Identification of Atrial Fibrillation With Single-Lead Mobile ECG During Normal Sinus Rhythm Using Deep Learning

BACKGROUND

The acquisition of single-lead electrocardiogram (ECG) from mobile devices offers a more practical approach to arrhythmia detection. Using artificial intelligence for atrial fibrillation (AF) identification enhances screening efficiency. However, the potential of single-lead ECG for AF identification during normal sinus rhythm (NSR) remains under-explored. This study introduces a method to identify AF using single-lead mobile ECG during NSR.

METHODS

We employed three deep learning models: recurrent neural network (RNN), long short-term memory (LSTM), and residual neural networks (ResNet50). From a dataset comprising 13,509 ECGs from 6,719 patients, 10,287 NSR ECGs from 5,170 patients were selected. Single-lead mobile ECGs underwent noise filtering and segmentation into 10-second intervals. A random under-sampling was applied to reduce bias from data imbalance. The final analysis involved 31,767 ECG segments, including 15,157 labeled as masked AF and 16,610 as Healthy.

RESULTS

ResNet50 outperformed the other models, achieving a recall of 79.3%, precision of 65.8%, F1-score of 71.9%, accuracy of 70.5%, and an area under the receiver operating characteristic curve (AUC) of 0.79 in identifying AF from NSR ECGs. Comparative performance scores for RNN and LSTM were 0.75 and 0.74, respectively. In an external validation set, ResNet50 attained an F1-score of 64.1%, recall of 68.9%, precision of 60.0%, accuracy of 63.4%, and AUC of 0.68.

CONCLUSION

The deep learning model using single-lead mobile ECG during NSR effectively identified AF at risk in future. However, further research is needed to enhance the performance of deep learning models for clinical application.

The evolving reality of digital health

Myriad digital health interventions, applications, devices and technologies have, and are, being developed to help refine and personalise medicine from the patient, healthcare professional (HCP), healthcare system and industry perspectives. At a gathering of leaders in digital health, discussion included the current landscape of such digital health tools (DHTs), with specific examples from cardiology and respiratory medicine, and both the benefits and sometime downfalls of such tools. While DHTs can help patients and HCPs detect and monitor health conditions, the experts discussed how adoption of DHTs may be hampered by issues such as access to technology; data privacy and security concerns; technology integration into current healthcare systems; cost and reimbursement; and lack of guidelines and regulatory hurdles. The experts suggested solutions to such issues, including wider availability of healthcare 'booths' local to a patient; easy to understand and use phone applications; patient and HCP incentives to use DHTs and clear paths to adoption within a healthcare system. These should help with integration of DHTs into the healthcare system to aid shared decision-making and, ultimately, streamline and personalise healthcare for all.

Classification of underlying paroxysmal supraventricular tachycardia types using deep learning of sinus rhythm electrocardiograms

Background

Obtaining tachycardia electrocardiograms (ECGs) in patients with paroxysmal supraventricular tachycardia (PSVT) is often challenging. Sinus rhythm ECGs are of limited predictive value for PSVT types in patients without preexcitation. This study aimed to explore the classification of atrioventricular nodal reentry tachycardia (AVNRT) and concealed atrioventricular reentry tachycardia (AVRT) using sinus rhythm ECGs through deep learning.

Methods

This retrospective study included patients diagnosed with either AVNRT or concealed AVRT, validated through electrophysiological studies. A modified ResNet-34 deep learning model, pre-trained on a public ECG database, was employed to classify sinus rhythm ECGs with underlying AVNRT or concealed AVRT. Various configurations were compared using ten-fold cross-validation on the training set, and the best-performing configuration was tested on the hold-out test set.

Results

The study analyzed 833 patients with AVNRT and 346 with concealed AVRT. Among ECG features, the corrected QT intervals exhibited the highest area under the receiver operating characteristic curve (AUROC) of 0.602. The performance of the deep learning model significantly improved after pre-training, showing an AUROC of 0.726 compared to 0.668 without pre-training (p < 0.001). No significant difference was found in AUROC between 12-lead and precordial 6-lead ECGs (p = 0.265). On the test set, deep learning achieved modest performance in differentiating the two types of arrhythmias, with an AUROC of 0.708, an AUPRC of 0.875, an F1-score of 0.750, a sensitivity of 0.670, and a specificity of 0.649.

Conclusion

The deep-learning classification of AVNRT and concealed AVRT using sinus rhythm ECGs is feasible, indicating potential for aiding in the non-invasive diagnosis of these arrhythmias.

Dynamic prediction of malignant ventricular arrhythmias using neural networks in patients with an implantable cardioverter-defibrillator

BACKGROUND

Risk stratification for ventricular arrhythmias currently relies on static measurements that fail to adequately capture dynamic interactions between arrhythmic substrate and triggers over time. We trained and internally validated a dynamic machine learning (ML) model and neural network that extracted features from longitudinally collected electrocardiograms (ECG), and used these to predict the risk of malignant ventricular arrhythmias.

METHODS

A multicentre study in patients implanted with an implantable cardioverter-defibrillator (ICD) between 2007 and 2021 in two academic hospitals was performed. Variational autoencoders (VAEs), which combine neural networks with variational inference principles, and can learn patterns and structure in data without explicit labelling, were trained to encode the mean ECG waveforms from the limb leads into 16 variables. Supervised dynamic ML models using these latent ECG representations and clinical baseline information were trained to predict malignant ventricular arrhythmias treated by the ICD. Model performance was evaluated on a hold-out set, using time-dependent receiver operating characteristic (ROC) and calibration curves.

FINDINGS

2942 patients (61.7 ± 13.9 years, 25.5% female) were included, with a total of 32,129 ECG recordings during a mean follow-up of 43.9 ± 35.9 months. The mean time-varying area under the ROC curve for the dynamic model was 0.738 ± 0.07, compared to 0.639 ± 0.03 for a static (i.e. baseline-only model). Feature analyses indicated dynamic changes in latent ECG representations, particularly those affecting the T-wave morphology, were of highest importance for model predictions.

INTERPRETATION

Dynamic ML models and neural networks effectively leverage routinely collected longitudinal ECG recordings for personalised and updated predictions of malignant ventricular arrhythmias, outperforming static models.

FUNDING

This publication is part of the project DEEP RISK ICD (with project number 452019308) of the research programme Rubicon which is (partly) financed by the Dutch Research Council (NWO). This research is partly funded by the Amsterdam Cardiovascular Sciences (personal grant F.V.Y.T).

Electrocardiogram Interpretation Using Artificial Intelligence: Diagnosis of Cardiac and Extracardiac Pathologic Conditions. How Far Has Machine Learning Reached?

Artificial intelligence (AI) is already widely used in different fields of medicine, making possible the integration of the paraclinical exams with the clinical findings in patients, for a more accurate and rapid diagnosis and treatment decision. The electrocardiogram remains one of the most important, fastest, cheapest, and noninvasive methods of diagnosis in cardiology, despite the rapid development and progression of the technology. Even if studied a long time ago, it still has a lot of less understood features that, with a better understanding, can give more clues to a correct and prompt diagnosis in a short time. The use of AI in the interpretation of the ECG improved the accuracy and the time to diagnosis in different cardiovascular diseases, and more than this, explaining the decision to make AI diagnosis improved the human understanding of the different features of the ECG that might be considered for a more accurate diagnosis. The purpose of this article is to provide an overview of the most recently published articles about the use of AI in ECG interpretation.

Deep Learning Models for Predicting Left Heart Abnormalities From Single-Lead Electrocardiogram for the Development of Wearable Devices

BACKGROUND

Left heart abnormalities are risk factors for heart failure. However, echocardiography is not always available. Electrocardiograms (ECGs), which are now available from wearable devices, have the potential to detect these abnormalities. Nevertheless, whether a model can detect left heart abnormalities from single Lead I ECG data remains unclear.

METHODS AND RESULTS

We developed Lead I ECG models to detect low ejection fraction (EF), wall motion abnormality, left ventricular hypertrophy (LVH), left ventricular dilatation, and left atrial dilatation. We used a dataset comprising 229,439 paired sets of ECG and echocardiography data from 8 facilities, and validated the model using external verification with data from 2 facilities. The area under the receiver operating characteristic curves of our model was 0.913 for low EF, 0.832 for wall motion abnormality, 0.797 for LVH, 0.838 for left ventricular dilatation, and 0.802 for left atrial dilatation. In interpretation tests with 12 cardiologists, the accuracy of the model was 78.3% for low EF and 68.3% for LVH. Compared with cardiologists who read the 12-lead ECGs, the model's performance was superior for LVH and similar for low EF.

CONCLUSIONS

From a multicenter study dataset, we developed models to predict left heart abnormalities using Lead I on the ECG. The Lead I ECG models show superior or equivalent performance to cardiologists using 12-lead ECGs.

Estimating age and gender from electrocardiogram signals: A comprehensive review of the past decade

Twelve lead electrocardiogram signals capture unique fingerprints about the body's biological processes and electrical activity of heart muscles. Machine learning and deep learning-based models can learn the embedded patterns in the electrocardiogram to estimate complex metrics such as age and gender that depend on multiple aspects of human physiology. ECG estimated age with respect to the chronological age reflects the overall well-being of the cardiovascular system, with significant positive deviations indicating an aged cardiovascular system and a higher likelihood of cardiovascular mortality. Several conventional, machine learning, and deep learning-based methods have been proposed to estimate age from electronic health records, health surveys, and ECG data. This manuscript comprehensively reviews the methodologies proposed for ECG-based age and gender estimation over the last decade. Specifically, the review highlights that elevated ECG age is associated with atherosclerotic cardiovascular disease, abnormal peripheral endothelial dysfunction, and high mortality, among many other cardiovascular disorders. Furthermore, the survey presents overarching observations and insights across methods for age and gender estimation. This paper also presents several essential methodological improvements and clinical applications of ECG-estimated age and gender to encourage further improvements of the state-of-the-art methodologies.

Prediction for blood lactate during exercise using an artificial intelligence-Enabled electrocardiogram: a feasibility study

Introduction: The acquisition of blood lactate concentration (BLC) during exercise is beneficial for endurance training, yet a convenient method to measure it remains unavailable. BLC and electrocardiogram (ECG) both exhibit variations with changes in exercise intensity and duration. In this study, we hypothesized that BLC during exercise can be predicted using ECG data. Methods: Thirty-one healthy participants underwent four cardiopulmonary exercise tests, including one incremental test and three constant work rate (CWR) tests at low, moderate, and high intensity. Venous blood samples were obtained immediately after each CWR test to measure BLC. A mathematical model was constructed using 31 trios of CWR tests, which utilized a residual network combined with long short-term memory to analyze every beat of lead II ECG waveform as 2D images. An artificial neural network was used to analyze variables such as the RR interval, age, sex, and body mass index. Results: The standard deviation of the fitting error was 0.12 mmol/L for low and moderate intensities, and 0.19 mmol/L for high intensity. Weighting analysis demonstrated that ECG data, including every beat of ECG waveform and RR interval, contribute predominantly. Conclusion: By employing 2D convolution and artificial neural network-based methods, BLC during exercise can be accurately estimated non-invasively using ECG data, which has potential applications in exercise training.

Completing the Cabrera Circle: deriving adaptable leads from ECG limb leads by combining constraints with a correction factor

Objective.We present a concept for processing 6-lead electrocardiography (ECG) signals which can be applied to various use cases in quantitative electrocardiography.Approach.Our work builds upon the mathematics of the well-known Cabrera sequence which is a re-sorting of the six limb leads (I,II,III,aVR,aVL,aVF) into a clockwise and physiologically-interpretable order. By deriving correction factors for harmonizing lead strengths and choosing an appropriate basis for the leads, we extend this concept towards what we call the 'Cabrera Circle' based on a mathematically sound foundation.Main results.To demonstrate the practical effectiveness and relevance of this concept, we analyze its suitability for deriving interpolated leads between the six limb leads and a 'radial' lead which both can be useful for specific use cases. We focus on the use cases of i) determination of the electrical heart axis by proposing a novel interactive tool for reconstructing the heart's vector loop and ii) improving accuracy in time of automatic R-wave detection and T-wave delineation in 6-lead ECG. For the first use case, we derive an equation which allows projections of the 2-dimensional vector loops to arbitrary angles of the Cabrera Circle. For the second use case, we apply several state-of-the-art algorithms to a freely-available 12-lead dataset (Lobachevsky University Database). Out-of-the-box results show that the derived radial lead outperforms the other limb leads (I,II,III,aVR,aVL,aVF) by improving F1 scores of R-peak and T-peak detection by 0.61 and 2.12, respectively. Results of on- and offset computations are also improved but on a smaller scale.Significance.In summary, the Cabrera Circle offers a methodology that might be useful for quantitative electrocardiography of the 6-lead subsystem-especially in the digital age.

Incorrect Electrocardiogram Lead Placement in ST-Segment-Elevation Myocardial Infarction

This case report describes a patient in their 70s with acute onset waxing and waning chest pressure, which radiated to both arms and was accompanied by shortness of breath.

Electrocardiographic Characteristics, Identification, and Management of Frequent Premature Ventricular Contractions

Premature ventricular complexes (PVCs) are frequently encountered in clinical practice. The association of PVCs with adverse cardiovascular outcomes is well established in the context of structural heart disease, yet not so much in the absence of structural heart disease. However, cardiac magnetic resonance (CMR) seems to contribute prognostically in the latter subgroup. PVC-induced myocardial dysfunction refers to the impairment of ventricular function due to PVCs and is mostly associated with a PVC burden > 10%. Surface 12-lead ECG has long been used to localize the anatomic site of origin and multiple algorithms have been developed to differentiate between right ventricular and left ventricular outflow tract (RVOT and LVOT, respectively) origin. Novel algorithms include alternative ECG lead configurations and, lately, sophisticated artificial intelligence methods have been utilized to determine the origins of outflow tract arrhythmias. The decision to therapeutically address PVCs should be made upon the presence of symptoms or the development of PVC-induced myocardial dysfunction. Therapeutic modalities include pharmacological therapy (I-C antiarrhythmic drugs and beta blockers), as well as catheter ablation, which has demonstrated superior efficacy and safety.

Generative adversarial networks in electrocardiogram synthesis: Recent developments and challenges

Training deep neural network classifiers for electrocardiograms (ECGs) requires sufficient data. However, imbalanced datasets pose a major problem for the training process and hence data augmentation is commonly performed. Generative adversarial networks (GANs) can create synthetic ECG data to augment such imbalanced datasets. This review aims at identifying the present literature concerning synthetic ECG signal generation using GANs to provide a comprehensive overview of architectures, quality evaluation metrics, and classification performances. Thirty publications from the years 2019 to 2022 were selected from three separate databases. Nine publications used a quality evaluation metric neglecting classification, eleven performed a classification but omitted a quality evaluation metric, and ten publications performed both. Twenty different quality evaluation metrics were observed. Overall, the classification performance of databases augmented with synthetically created ECG signals increased by 7 % to 98 % in accuracy and 6 % to 97 % in sensitivity. In conclusion, synthetic ECG signal generation using GANs represents a promising tool for data augmentation of imbalanced datasets. Consistent quality evaluation of generated signals remains challenging. Hence, future work should focus on the establishment of a gold standard for quality evaluation metrics for GANs.

Improving detection of obstructive coronary artery disease with an artificial intelligence-enabled electrocardiogram algorithm

BACKGROUND AND AIMS

To evaluate the risk of coronary artery disease (CAD), the traditional approach involves assessing the patient's symptoms, traditional cardiovascular risk factors (CVRFs), and a 12-lead electrocardiogram (ECG). However, currently, there are no established criteria for interpreting an ECG to diagnose CAD. Therefore, we sought to develop an artificial intelligence (AI)-enabled ECG model to assist in identifying patients with CAD.

METHODS

In this study, we included patients who underwent coronary angiography (CAG) at a single center between 2017 and 2019. Preprocedural 12-lead ECG performed within 24 h was obtained. Obstructive CAD was defined as ≥ 50% diameter stenosis. Using age, gender and ECG data, we developed stacking models using both deep learning and machine learning. Then we compared the performance of our models with CVRFs and with cardiologists' ECG interpretation. Additionally, we validated our model on an external cohort from a different hospital.

RESULTS

We included 4951 patients with a mean age of 65.5 ± 12.5 years, of whom 67.0% were men. Based on CAG, obstructive CAD was confirmed in 2637 patients (53.2%). Our best AI model demonstrated comparable performance to CVRFs in predicting CAD, with an AUC of 0.70 (95% CI: 0.66-0.75) compared to 0.71 (95% CI: 0.66-0.76). The sensitivity and specificity of the AI model were 0.75 and 0.54, respectively, while those of CVRFs were 0.67 and 0.63. Compared to cardiologists, the AI model showed better performance with an F1 score of 0.68 vs 0.41. The external validation showed generally consistent diagnostic findings, although there was a slightly lower level of agreement observed in the external cohort. Incorporating ECG and CVRFs improved the AUC to 0.72.

CONCLUSIONS

Our study suggests that an AI-enabled ECG model can assist in identifying patients with obstructive CAD, with diagnostic performance similar to that of the traditional approach based on CVRFs. This model could serve as a useful clinical tool in an outpatient setting to identify patients who require further diagnostic tests.

Deep learning-based model detects atrial septal defects from electrocardiography: a cross-sectional multicenter hospital-based study

Background

Atrial septal defect (ASD) increases the risk of adverse cardiovascular outcomes. Despite the potential for risk mitigation through minimally invasive percutaneous closure, ASD remains underdiagnosed due to subtle symptoms and examination findings. To bridge this diagnostic gap, we propose a novel screening strategy aimed at early detection and enhanced diagnosis through the implementation of a convolutional neural network (CNN) to identify ASD from 12-lead electrocardiography (ECG).

Methods

ECGs were collected from patients with at least one recorded echocardiogram at 3 hospitals from 2 continents (Keio University Hospital from July 2011 to December 2020, Brigham and Women's Hospital from January 2015 to December 2020, and Dokkyo Medical University Saitama Medical Center from January 2010 and December 2021). ECGs from patients with a diagnosis of ASD were labeled as positive cases while the remainder were labeled as negative. ECGs after the closure of ASD were excluded. After randomly splitting the ECGs into 3 datasets (50% derivation, 20% validation, and 30% test) with no patient overlap, a CNN-based model was trained using the derivation datasets from 2 hospitals and was tested on held-out datasets along with an external validation on the 3rd hospital. All eligible ECGs were used for derivation and validation whereas the earliest ECG for each patient was used for the test and external validation. The discrimination of ASD was assessed by the area under the receiver operating characteristic curve (AUROC). Multiple subgroups were examined to identify any heterogeneity.

Findings

A total of 671,201 ECGs from 80,947 patients were collected from the 3 institutions. The AUROC for detecting ASD was 0.85-0.90 across the 3 hospitals. The subgroup analysis showed excellent performance across various characteristics Screening simulation using the model greatly increased sensitivity from 80.6% to 93.7% at specificity 33.6% when compared to using overt ECG abnormalities.

Interpretation

A CNN-based model using 12-lead ECG successfully identified the presence of ASD with excellent generalizability across institutions from 2 separate continents.

Funding

This work was supported by research grants from JST (JPMJPF2101), JSR corporation, Taiju Life Social Welfare Foundation, Kondou Kinen Medical Foundation, Research fund of Mitsukoshi health and welfare foundation, Tokai University School of Medicine Project Research and Internal Medicine Project Research, Secom Science and Technology Foundation, and Grants from AMED (JP23hma922012 and JP23ym0126813). This work was partially supported by One Brave Idea, co-funded by the American Heart Association and Verily with significant support from AstraZeneca and pillar support from Quest Diagnostics.

The digital journey: 25 years of digital development in electrophysiology from an Europace perspective

AIMS

Over the past 25 years there has been a substantial development in the field of digital electrophysiology (EP) and in parallel a substantial increase in publications on digital cardiology.In this celebratory paper, we provide an overview of the digital field by highlighting publications from the field focusing on the EP Europace journal.

RESULTS

In this journey across the past quarter of a century we follow the development of digital tools commonly used in the clinic spanning from the initiation of digital clinics through the early days of telemonitoring, to wearables, mobile applications, and the use of fully virtual clinics. We then provide a chronicle of the field of artificial intelligence, a regulatory perspective, and at the end of our journey provide a future outlook for digital EP.

CONCLUSION

Over the past 25 years Europace has published a substantial number of papers on digital EP, with a marked expansion in digital publications in recent years.

Development and validation of a rapid visual technique for left ventricular hypertrophy detection from the electrocardiogram

Introduction

Left ventricular hypertrophy (LVH) detection techniques on by electrocardiogram (ECG) are cumbersome to remember with modest performance. This study validated a rapid technique for LVH detection and measured its performance against other techniques.

Methods

This was a retrospective cohort study of patients at Stanford Health Care who received ECGs and resting transthoracic echocardiograms (TTE) from 2006 through 2018. The novel technique, Witteles-Somani (WS), assesses for S- and R-wave overlap on adjacent precordial leads. The WS, Sokolow-Lyon, Cornell, and Peguero-Lo Presti techniques were algorithmically implemented on ECGs. Classification metrics, receiver-operator curves, and Pearson correlations measured performance. Age- and sex-adjusted Cox proportional hazard models evaluated associations between incident cardiovascular outcomes and each technique.

Results

A total of 53,333 ECG-TTE pairs from 18,873 patients were identified. Of all ECG-TTE pairs, 21,638 (40.6%) had TTE-diagnosed LVH. The WS technique had a sensitivity of 0.46, specificity of 0.66, and AUROC of 0.56, compared to Sokolow-Lyon (AUROC 0.55), Cornell (AUROC 0.63), and Peguero-Lo Presti (AUROC 0.63). Patients meeting LVH by WS technique had a higher risk of cardiovascular mortality [HR 1.18, 95% CI (1.12, 1.24), P < 0.001] and a higher risk of developing any cardiovascular disease [HR 1.29, 95% CI (1.22, 1.36), P < 0.001], myocardial infarction [HR 1.60, 95% CI (1.44, 1.78), P < 0.005], and heart failure [HR 1.24, 95% CI (1.17, 1.32), P < 0.001].

Conclusions

The WS criteria is a rapid visual technique for LVH detection with performance like other LVH detection techniques and is associated with incident cardiovascular outcomes.

A Novel ECG-Based Deep Learning Algorithm to Predict Cardiomyopathy in Patients With Premature Ventricular Complexes

BACKGROUND

Premature ventricular complexes (PVCs) are prevalent and, although often benign, they may lead to PVC-induced cardiomyopathy. We created a deep-learning algorithm to predict left ventricular ejection fraction (LVEF) reduction in patients with PVCs from a 12-lead electrocardiogram (ECG).

OBJECTIVES

This study aims to assess a deep-learning model to predict cardiomyopathy among patients with PVCs.

METHODS

We used electronic medical records from 5 hospitals and identified ECGs from adults with documented PVCs. Internal training and testing were performed at one hospital. External validation was performed with the others. The primary outcome was first diagnosis of LVEF ≤40% within 6 months. The dataset included 383,514 ECGs, of which 14,241 remained for analysis. We analyzed area under the receiver operating curves and explainability plots for representative patients, algorithm prediction, PVC burden, and demographics in a multivariable Cox model to assess independent predictors for cardiomyopathy.

RESULTS

Among the 14,241-patient cohort (age 67.6 ± 14.8 years; female 43.8%; White 29.5%, Black 8.6%, Hispanic 6.5%, Asian 2.2%), 22.9% experienced reductions in LVEF to ≤40% within 6 months. The model predicted reductions in LVEF to ≤40% with area under the receiver operating curve of 0.79 (95% CI: 0.77-0.81). The gradient weighted class activation map explainability framework highlighted the sinus rhythm QRS complex-ST segment. In patients who underwent successful PVC ablation there was a post-ablation improvement in LVEF with resolution of cardiomyopathy in most (89%) patients.

CONCLUSIONS

Deep-learning on the 12-lead ECG alone can accurately predict new-onset cardiomyopathy in patients with PVCs independent of PVC burden. Model prediction performed well across sex and race, relying on the QRS complex/ST-segment in sinus rhythm, not PVC morphology.

Comparison of discrimination and calibration performance of ECG-based machine learning models for prediction of new-onset atrial fibrillation

BACKGROUND

Machine learning (ML) methods to build prediction models starting from electrocardiogram (ECG) signals are an emerging research field. The aim of the present study is to investigate the performances of two ML approaches based on ECGs for the prediction of new-onset atrial fibrillation (AF), in terms of discrimination, calibration and sample size dependence.

METHODS

We trained two models to predict new-onset AF: a convolutional neural network (CNN), that takes as input the raw ECG signals, and an eXtreme Gradient Boosting model (XGB), that uses the signal's extracted features. A penalized logistic regression model (LR) was used as a benchmark. Discrimination was evaluated with the area under the ROC curve, while calibration with the integrated calibration index. We investigated the dependence of models' performances on the sample size and on class imbalance corrections introduced with random under-sampling.

RESULTS

CNN's discrimination was the most affected by the sample size, outperforming XGB and LR only around n = 10.000 observations. Calibration showed only a small dependence on the sample size for all the models considered. Balancing the training set with random undersampling did not improve discrimination in any of the models. Instead, the main effect of imbalance corrections was to worsen the models' calibration (for CNN, integrated calibration index from 0.014 [0.01, 0.018] to 0.17 [0.16, 0.19]). The sample size emerged as a fundamental point for developing the CNN model, especially in terms of discrimination (AUC = 0.75 [0.73, 0.77] when n = 10.000, AUC = 0.80 [0.79, 0.81] when n = 150.000). The effect of the sample size on the other two models was weaker. Imbalance corrections led to poorly calibrated models, for all the approaches considered, reducing the clinical utility of the models.

CONCLUSIONS

Our results suggest that the choice of approach in the analysis of ECG should be based on the amount of data available, preferring more standard models for small datasets. Moreover, imbalance correction methods should be avoided when developing clinical prediction models, where calibration is crucial.

Artificial Deep Neural Network for Sensorless Pump Flow and Hemodynamics Estimation During Continuous-Flow Mechanical Circulatory Support

The objective of this study was to compare the estimates of pump flow and systemic vascular resistance (SVR) derived from a mathematical regression model to those from an artificial deep neural network (ADNN). Hemodynamic and pump-related data were generated using both the Cleveland Clinic continuous-flow total artificial heart (CFTAH) and pediatric CFTAH on a mock circulatory loop. An ADNN was trained with generated data, and a mathematical regression model was also generated using the same data. Finally, the absolute error for the actual measured data and each set of estimated data were compared. A strong correlation was observed between the measured flow and the estimated flow using either method (mathematical, R = 0.97, p < 0.01; ADNN, R = 0.99, p < 0.01). The absolute error was smaller in the ADNN estimation (mathematical, 0.3 L/min; ADNN 0.12 L/min; p < 0.01). Furthermore, strong correlation was observed between measured and estimated SVR (mathematical, R = 0.97, p < 0.01; ADNN, R = 0.99, p < 0.01). The absolute error for ADNN estimation was also smaller than that of the mathematical estimation (mathematical, 463 dynes·sec·cm -5 ; ADNN, 123 dynes·sec·cm -5 , p < 0.01). Therefore, in this study, ADNN estimation was more accurate than mathematical regression estimation. http://links.lww.com/ASAIO/A991.

The role of artificial intelligence in hypertensive disorders of pregnancy: towards personalized healthcare

INTRODUCTION

Guidelines advise ongoing follow-up of patients after hypertensive disorders of pregnancy (HDP) to assess cardiovascular risk and manage future patient-specific pregnancy conditions. However, there are limited tools available to monitor patients, with those available tending to be simple risk assessments that lack personalization. A promising approach could be the emerging artificial intelligence (AI)-based techniques, developed from big patient datasets to provide personalized recommendations for preventive advice.

AREAS COVERED

In this narrative review, we discuss the impact of integrating AI and big data analysis for personalized cardiovascular care, focusing on the management of HDP.

EXPERT OPINION

The pathophysiological response of women to pregnancy varies, and deeper insight into each response can be gained through a deeper analysis of the medical history of pregnant women based on clinical records and imaging data. Further research is required to be able to implement AI for clinical cases using multi-modality and multi-organ assessment, and this could expand both knowledge on pregnancy-related disorders and personalized treatment planning.

Artificial intelligence-enabled tools in cardiovascular medicine: A survey of current use, perceptions, and challenges

Background

Numerous artificial intelligence (AI)-enabled tools for cardiovascular diseases have been published, with a high impact on public health. However, few have been adopted into, or have meaningfully affected, routine clinical care.

Objective

To evaluate current awareness, perceptions, and clinical use of AI-enabled digital health tools for patients with cardiovascular disease, and challenges to adoption.

Methods

This mixed-methods study included interviews with 12 cardiologists and 8 health information technology (IT) administrators, and a follow-on survey of 90 cardiologists and 30 IT administrators.

Results

We identified 5 major challenges: (1) limited knowledge, (2) insufficient usability, (3) cost constraints, (4) poor electronic health record interoperability, and (5) lack of trust. A minority of cardiologists were using AI tools; more were prepared to implement AI tools, but their sophistication level varied greatly.

Conclusion

Most respondents believe in the potential of AI-enabled tools to improve care quality and efficiency, but they identified several fundamental barriers to wide-scale adoption.

Cardiovascular imaging in conduction system pacing: What does the clinician need?

Permanent pacemakers are used for symptomatic bradycardia and biventricular pacing (BVP)-cardiac resynchronization therapy (BVP-CRT) is established for heart failure (HF) patients traditionally. According to guidelines, patients' selection for CRT is based on QRS duration (QRSd) and morphology by surface electrocardiogram (ECG). Cardiovascular imaging techniques evaluate cardiac structure and function as well as identify pathophysiological substrate changes including the presence of scar. Cardiovascular imaging helps by improving the selection of candidates, guiding left ventricular (LV) lead placement, and optimization devices during the follow-up. Conduction system pacing (CSP) includes His bundle pacing (HBP) and left bundle branch pacing (LBBP) which is screwed into the interventricular septum. CSP maintains and restores ventricular synchrony in patients with native narrow QRSd and left bundle branch block (LBBB), respectively. LBBP is more feasible than HBP due to a wider target area. This review highlights the role of multimodality cardiovascular imaging including fluoroscopy, echocardiography, cardiac magnetic resonance (CMR), myocardial scintigraphy, and computed tomography (CT) in the pre-procedure assessment for CSP, better selection for CSP candidates, the guidance of CSP lead implantation, and the optimization of devices programming after the procedure. We also compare the different characteristics of multimodality imaging and discuss their potential roles in future CSP implantation.

DeepMiceTL: a deep transfer learning based prediction of mice cardiac conduction diseases using early electrocardiograms

Cardiac conduction disease is a major cause of morbidity and mortality worldwide. There is considerable clinical significance and an emerging need of early detection of these diseases for preventive treatment success before more severe arrhythmias occur. However, developing such early screening tools is challenging due to the lack of early electrocardiograms (ECGs) before symptoms occur in patients. Mouse models are widely used in cardiac arrhythmia research. The goal of this paper is to develop deep learning models to predict cardiac conduction diseases in mice using their early ECGs. We hypothesize that mutant mice present subtle abnormalities in their early ECGs before severe arrhythmias present. These subtle patterns can be detected by deep learning though they are hard to be identified by human eyes. We propose a deep transfer learning model, DeepMiceTL, which leverages knowledge from human ECGs to learn mouse ECG patterns. We further apply the Bayesian optimization and $k$-fold cross validation methods to tune the hyperparameters of the DeepMiceTL. Our results show that DeepMiceTL achieves a promising performance (F1-score: 83.8%, accuracy: 84.8%) in predicting the occurrence of cardiac conduction diseases using early mouse ECGs. This study is among the first efforts that use state-of-the-art deep transfer learning to identify ECG patterns during the early course of cardiac conduction disease in mice. Our approach not only could help in cardiac conduction disease research in mice, but also suggest a feasibility for early clinical diagnosis of human cardiac conduction diseases and other types of cardiac arrythmias using deep transfer learning in the future.

Automated diagnosis of cardiovascular diseases from cardiac magnetic resonance imaging using deep learning models: A review

In recent years, cardiovascular diseases (CVDs) have become one of the leading causes of mortality globally. At early stages, CVDs appear with minor symptoms and progressively get worse. The majority of people experience symptoms such as exhaustion, shortness of breath, ankle swelling, fluid retention, and other symptoms when starting CVD. Coronary artery disease (CAD), arrhythmia, cardiomyopathy, congenital heart defect (CHD), mitral regurgitation, and angina are the most common CVDs. Clinical methods such as blood tests, electrocardiography (ECG) signals, and medical imaging are the most effective methods used for the detection of CVDs. Among the diagnostic methods, cardiac magnetic resonance imaging (CMRI) is increasingly used to diagnose, monitor the disease, plan treatment and predict CVDs. Coupled with all the advantages of CMR data, CVDs diagnosis is challenging for physicians as each scan has many slices of data, and the contrast of it might be low. To address these issues, deep learning (DL) techniques have been employed in the diagnosis of CVDs using CMR data, and much research is currently being conducted in this field. This review provides an overview of the studies performed in CVDs detection using CMR images and DL techniques. The introduction section examined CVDs types, diagnostic methods, and the most important medical imaging techniques. The following presents research to detect CVDs using CMR images and the most significant DL methods. Another section discussed the challenges in diagnosing CVDs from CMRI data. Next, the discussion section discusses the results of this review, and future work in CVDs diagnosis from CMR images and DL techniques are outlined. Finally, the most important findings of this study are presented in the conclusion section.

Automatic screening of patients with atrial fibrillation from 24-h Holter recording using deep learning

Aims

As the demand for atrial fibrillation (AF) screening increases, clinicians spend a significant amount of time identifying AF signals from massive amounts of data obtained during long-term dynamic electrocardiogram (ECG) monitoring. The identification of AF signals is subjective and depends on the experience of clinicians. However, experienced cardiologists are scarce. This study aimed to apply a deep learning-based algorithm to fully automate primary screening of patients with AF using 24-h Holter monitoring.

Methods and results

A deep learning model was developed to automatically detect AF episodes using RR intervals and was trained and evaluated on 23 621 (2297 AF and 21 324 non-AF) 24-h Holter recordings from 23 452 patients. Based on the AF episode detection results, patients with AF were automatically identified using the criterion of at least one AF episode lasting 6 min or longer. Performance was assessed on an independent real-world hospital-scenario test set (19 227 recordings) and a community-scenario test set (1299 recordings). For the two test sets, the model obtained high performance for the identification of patients with AF (sensitivity: 0.995 and 1.000; specificity: 0.985 and 0.997, respectively). Moreover, it obtained good and consistent performance (sensitivity: 1.000; specificity: 0.972) for an external public data set.

Conclusion

Using the criterion of at least one AF episode of 6 min or longer, the deep learning model can fully automatically screen patients for AF with high accuracy from long-term Holter monitoring data. This method may serve as a powerful and cost-effective tool for primary screening for AF.

The feasibility of early detecting coronary artery disease using deep learning-based algorithm based on electrocardiography

BACKGROUND

Coronary Artery Disease (CAD) is a major cause of morbidity and mortality, yet it is frequently asymptomatic in the early stages and hence goes undetected.

OBJECTIVE

We aimed to develop a novel artificial intelligence-based approach for early detection of CAD patients based solely on electrocardiogram (ECG).

METHODS

This study included patients with suspected CAD who had standard 10-s resting 12-lead ECGs and coronary computed tomography angiography (cCTA) results within 4 weeks or less. The ECG and cCTA data from the same patient were matched based on their hospitalization or outpatient ID. All matched data pairs were then randomly divided into training, validation dataset for model development based on convolutional neural network (CNN) and test dataset for model evaluation. The accuracy (Acc), specificity (Spec), sensitivity (Sen), positive predictive value (PPV), negative predictive value (NPV) and area under the receiver operating characteristic curve (AUC) of the model were calculated by using the test dataset.

RESULTS

In the test dataset, the model for detecting CAD achieved an AUC of 0.75 (95% CI, 0.73 to 0.78) with an accuracy of 70.0%. Using the optimal cut-off point, the CAD detection model had sensitivity of 68.7%, specificity of 70.9%, positive predictive value (PPV) of 61.2%, and negative predictive value (NPV) of 77.2%. Our study demonstrates that a well-trained CNN model based solely on ECG could be considered an efficient, low-cost, and noninvasive method of assisting in CAD detection.

Artificial Intelligence in Cardiology: Applications and Obstacles

Artificial intelligence (AI) technology is poised to alter the flow of daily life, and in particular, medicine, where it may eventually complement the physician's work in diagnosing and treating disease. Despite the recent frenzy and uptick in AI research over the past decade, the integration of AI into medical practice is in its early stages. Cardiology stands to benefit due to its many diagnostic modalities and diverse treatments. AI methods have been applied to various domains within cardiology: imaging, electrocardiography, wearable devices, risk prediction, and disease classification. While many AI-based approaches have been developed that perform equal to or better than the state-of-the-art, few prospective randomized studies have evaluated their use. Furthermore, obstacles at the intersection of medicine and AI remain unsolved, including model understanding, bias, model evaluation, relevance and reproducibility, and legal and ethical dilemmas. We summarize recent and current applications of AI in cardiology, followed by a discussion of the aforementioned complications.