Why the study?
AI-based models show promise for detecting cardiovascular diseases from ECGs, but external validation is lacking for many published algorithms, including an existing LVSD detection model.
Does an AI-based model applied to 12-lead ECGs accurately detect left ventricular systolic dysfunction compared to echocardiography?
Population
42 291 ECG-echocardiography pairs from patients at Heart Center Leipzig
Comparison
AI-based model ECG probability for LVSD vs echocardiography reference standard
Design
Retrospective external validation study
Follow-up
≥3 months
Key result
An artificial intelligence-based model accurately detected left ventricular systolic dysfunction from 12-lead ECGs with an AUROC of 0.88, 82% sensitivity, and 77% specificity.
Authors
Loading...
May aid LVSD detection on routine ECGs; leaves open whether implementation improves outcomes.
Observational (n=42,291)
No
Does an AI-based model applied to 12-lead ECGs accurately detect left ventricular systolic dysfunction compared to echocardiography?
Effect estimate: AUROC 0.88
An AI-based model applied to standard 12-lead ECGs can accurately detect left ventricular systolic dysfunction, and high-probability false positives may predict future development of the condition.
König et al. (2023) conducted an observational in Left ventricular systolic dysfunction (n=42,291). Artificial intelligence-based model for 12-lead ECGs vs. Echocardiography was evaluated on Detection of left ventricular systolic dysfunction (AUROC 0.88). An artificial intelligence-based model accurately detected left ventricular systolic dysfunction from 12-lead ECGs with an AUROC of 0.88, 82% sensitivity, and 77% specificity.