Comparison of Autoencoder Encodings for ECG Representation in Downstream Prediction Tasks

TL;DR

This paper compares various autoencoder-based methods for ECG data encoding, demonstrating that novel VAE variants can effectively reduce data complexity and improve downstream prediction accuracy with less data and computational effort.

Contribution

It introduces three new VAE variants and evaluates their effectiveness in ECG encoding, showing improved signal reconstruction and prediction performance over traditional methods.

Findings

Abeta-VAE achieved MAE of 15.7 microvolts, comparable to signal noise.

SAE encodings improved LVEF prediction with AUROC of 0.901.

VAE encodings enable effective ECG analysis with limited data.

Abstract

The electrocardiogram (ECG) is an inexpensive and widely available tool for cardiovascular assessment. Despite its standardized format and small file size, the high complexity and inter-individual variability of ECG signals (typically a 60,000-size vector) make it challenging to use in deep learning models, especially when only small datasets are available. This study addresses these challenges by exploring feature generation methods from representative beat ECGs, focusing on Principal Component Analysis (PCA) and Autoencoders to reduce data complexity. We introduce three novel Variational Autoencoder (VAE) variants: Stochastic Autoencoder (SAE), Annealed beta-VAE (Abeta-VAE), and cyclical beta-VAE (Cbeta-VAE), and compare their effectiveness in maintaining signal fidelity and enhancing downstream prediction tasks. The Abeta-VAE achieved superior signal reconstruction, reducing the mean absolute error (MAE) to 15.7 plus-minus 3.2 microvolts, which is at the level of signal noise. Moreover, the SAE encodings, when combined with ECG summary features, improved the prediction of reduced Left Ventricular Ejection Fraction (LVEF), achieving an area under the receiver operating characteristic curve (AUROC) of 0.901. This performance nearly matches the 0.910 AUROC of state-of-the-art CNN models but requires significantly less data and computational resources. Our findings demonstrate that these VAE encodings are not only effective in simplifying ECG data but also provide a practical solution for applying deep learning in contexts with limited-scale labeled training data.