English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We propose 3KG, a physiologically-inspired contrastive learning approach thatgenerates views using 3D augmentations of the 12-lead electrocardiogram. Weevaluate representation quality by fine-tuning a linear layer for thedownstream task of 23-class diagnosis on the PhysioNet 2020 challenge trainingdata and find that 3KG achieves a $9.1\%$ increase in mean AUC over the bestself-supervised baseline when trained on $1\%$ of labeled data. Our empiricalanalysis shows that combining spatial and temporal augmentations produces thestrongest representations. In addition, we investigate the effect of thisphysiologically-inspired pretraining on downstream performance on differentdisease subgroups and find that 3KG makes the greatest gains for conduction andrhythm abnormalities. Our method allows for flexibility in incorporating otherself-supervised strategies and highlights the potential for similarmodality-specific augmentations for other biomedical signals.

3KG: Contrastive Learning of 12-Lead Electrocardiograms using Physiologically-Inspired Augmentations