GeoECG: Data Augmentation via Wasserstein Geodesic Perturbation for Robust Electrocardiogram Prediction

Abstract

There has been an increased interest in applying deep neural networks to automatically interpret and analyze the 12-lead electrocardiogram (ECG). The current paradigms with machine learning methods are often limited by the amount of labeled data. This phenomenon is particularly problematic for clinically-relevant data, where labeling at scale can be time-consuming and costly in terms of the specialized expertise and human effort required. Moreover, deep learning classifiers may be vulnerable to adversarial examples and perturbations, which could have catastrophic consequences, for example, when applied in the context of medical treatment, clinical trials, or insurance claims. In this paper, we propose a physiologically-inspired data augmentation method to improve performance and increase the robustness of heart disease detection based on ECG signals. We obtain augmented samples by perturbing the data distribution towards other classes along the geodesic in Wasserstein space. To better utilize domain-specific knowledge, we design a

A Systematic Survey of Data Augmentation of ECG Signals for AI Applications

“…The most common implementation of AI in ECG analysis is automatic ECG interpretation [14,15,17,18]. Other important applications include localizing and annotating specific rhythms and beats, which can aid in the detection of conditions such as MI [13] and fetal heart rate series classification [19]. Moreover, recent advancements in biometricbased human identification show great promise for accurate recognition based on ECG data [20,21].…”

“…Silva et al [12] designed a cardiorespiratory signal synthesizer by conditional sampling from a multimodally trained stochastic system of Gaussian copulas integrated with an MC. Zhu et al [13] proposed a novel DA technique that took into account both probability distribution and geometry. In their technique, they introduced variations to the data distribution along the geodesic in a Wasserstein space, which is a mathematical concept used to measure the distance between two probability distributions.…”

“…Advanced DA techniques follow the distributions of features present in the dataset to produce novel pattern combinations. Many statistical models, such as the Gaussian mixture model (GMM) and Markov chain (MC) models have been proposed for generating new samples [11][12][13]. Learning-based models such as variational autoencoder (VAE) [14] and generative adversarial networks (GANs) [15], are relatively new models developed recently for the generation of ECGs.…”