Aims This study aims to assess whether information derived from the raw 12-lead electrocardiogram (ECG) combined with clinical information is predictive of atrial fibrillation (AF) development. Methods We use a subset of the Telehealth Network of Minas Gerais (TNMG) database consisting of patients that had repeated 12-lead ECG measurements between 2010-2017 that is 1,130,404 recordings from 415,389 unique patients. Median and interquartile of age for the recordings were 58 (46-69) and 38% of the patients were males. Recordings were assigned to train-validation and test sets in an 80:20% split which was stratified by class, age and gender. A random forest classifier was trained to predict, for a given recording, the risk of AF development within 5-years. We use features obtained from different modalities, namely demographics, clinical information, engineered features, and features from deep representation learning. Results The best model performance on the test set was obtained for the model combining features from all modalities with an AUROC=0.909 against the best single modality model which had an AUROC=0.839. Conclusion Our study has important clinical implications for AF management. It is the first study integrating feature engineering, deep learning and EMR metadata to create a risk prediction tool for the management of patients at risk of AF. The best model that includes features from all modalities demonstrates that human knowledge in electrophysiology combined with deep learning outperforms any single modality approach. The high performance obtained suggest that structural changes in the 12-lead ECG are associated with existing or impending AF.

No abstract

No abstract

To drive health innovation that meets the needs of all and democratize healthcare, there is a need to assess the generalization performance of deep learning (DL) algorithms across various distribution shifts to ensure that these algorithms are robust. This retrospective study is, to the best of our knowledge, an original attempt to develop and assess the generalization performance of a DL model for AF events detection from long term beat-to-beat intervals across geography, ages and sexes. The new recurrent DL model, denoted ArNet2, is developed on a large retrospective dataset of 2,147 patients totaling 51,386 h obtained from continuous electrocardiogram (ECG). The model’s generalization is evaluated on manually annotated test sets from four centers (USA, Israel, Japan and China) totaling 402 patients. The model is further validated on a retrospective dataset of 1,825 consecutives Holter recordings from Israel. The model outperforms benchmark state-of-the-art models and generalized well across geography, ages and sexes. For the task of event detection ArNet2 performance was higher for female than male, higher for young adults (less than 61 years old) than other age groups and across geography. Finally, ArNet2 shows better performance for the test sets from the USA and China. The main finding explaining these variations is an impairment in performance in groups with a higher prevalence of atrial flutter (AFL). Our findings on the relative performance of ArNet2 across groups may have clinical implications on the choice of the preferred AF examination method to use relative to the group of interest.

Objective. Arrhythmia is an abnormal cardiac rhythm that affects the pattern and rate of the heartbeat. Wearable devices with the functionality to measure and store heart rate (HR) data are growing in popularity and enable diagnosing and monitoring arrhythmia on a large scale. The typical sampling resolution of HR data available from non-medical grade wearable devices varies from seconds to several minutes depending on the device and its settings. However, the impact of sampling resolution on the performance and quality of arrhythmia detection has not yet been quantified. Approach. In this study, we investigated the detection and classification of three arrhythmias, namely atrial fibrillation, bradycardia, tachycardia, from down-sampled HR data with various temporal resolution (5-, 15-, 30- and 60 s averages) in 1 h segments extracted from an annotated Holter ECG database acquired at the University of Virginia Heart Station. For the classification task, a total of 15 common heart rate variability (HRV) features were engineered based on the HR time series of each patient. Three different types of machine learning classifiers were evaluated, namely logistic regression, support vector machine and random forest. Main results. A decrease in temporal resolution drastically impacted the detection of atrial fibrillation but did not substantially affect the detection of bradycardia and tachycardia. A HR resolution up to 15 s average demonstrated reasonable performance with a sensitivity of 0.92 and a specificity of 0.86 for a multiclass random forest classifier. Significance. HRV features extracted from low resolution long HR recordings have the potential to increase the early detection of arrhythmias in undiagnosed individuals.