Key result
A variational autoencoder generated realistic synthetic ECG signals for a single cardiac cycle using 25 interpretable features, achieving a Maximum Mean Discrepancy metric of 3.83 × 10^-3.
Why the study?
The study was conducted to encode ECG signals using as few features as possible to address the lack of labeled ECG data for supervised learning.
A variational autoencoder can generate realistic synthetic ECG cardiac cycles and extract 25 interpretable features, which may aid in training supervised learning models for cardiovascular diagnostics.
We propose a method for generating an electrocardiogram (ECG) signal for one cardiac cycle using a variational autoencoder. Our goal was to encode the original ECG signal using as few features as possible. Using this method we extracted a vector of new 25 features, which in many cases can be interpreted. The generated ECG has quite natural appearance. The low value of the Maximum Mean Discrepancy metric, 3.83 × 10 −3 , indicates good quality of ECG generation too. The extracted new features will help to improve the quality of automatic diagnostics of cardiovascular diseases. Generating new synthetic ECGs will allow us to solve the issue of the lack of labeled ECG for using them in supervised learning.
No takes yet. Share an insight, caveat, or question.
Kuznetsov et al. (2021) studied Electrocardiogram (ECG) generation (n=2,033). Variational Autoencoder (VAE) was evaluated on Maximum Mean Discrepancy (MMD) metric. A variational autoencoder generated realistic synthetic ECG signals for a single cardiac cycle using 25 interpretable features, achieving a Maximum Mean Discrepancy metric of 3.83 × 10^-3.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: