Key result
A deep learning model classified atrial fibrillation in critically ill patients using continuous telemetry data with an overall sensitivity of 84%, specificity of 89%, and an AUROC of 0.935.
Why the study?
New-onset AF in the intensive care unit is often paroxysmal and fleeting, making it difficult to diagnose and quantify, prompting the need for automated detection algorithms.
Does a deep learning model accurately classify atrial fibrillation in critically ill patients using continuous telemetry data?
Population
984 critically ill patients with continuous telemetry data
Comparison
Deep learning model classification stratified by clean vs noisy signal quality
Design
Model development and external validation study
Authors
Loading...
May enhance NOAF burden quantification in ICU; leaves open prospective validation of automated algorithms.
Observational (n=984)
No
Does a deep learning model accurately classify atrial fibrillation in critically ill patients using continuous telemetry data?
Effect estimate: AUROC 0.935
A deep learning model trained on static ECGs can effectively detect atrial fibrillation in continuous ICU telemetry data, offering a scalable method to quantify AF burden in critically ill patients.
Chen et al. (2023) conducted an observational in Atrial fibrillation (n=984). Deep learning model for AF classification vs. Expert annotation (ground truth) was evaluated on Classification of atrial fibrillation (AUROC) (AUROC 0.935). A deep learning model classified atrial fibrillation in critically ill patients using continuous telemetry data with an overall sensitivity of 84%, specificity of 89%, and an AUROC of 0.935.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: