Why the study?
AI-enabled ECG models for LVSD screening dropped unexpectedly in external validation, prompting the need to evaluate de novo models across multiple institutions and patient strata.
Does an AI-enabled 12-lead ECG model accurately detect ejection fraction <40% across different institutions and patient strata?
Population
221 846 ECGs taken within 14 days of echocardiography across four academic hospitals
Comparison
AI models evaluated across internal held-out test data sets vs three external institutions
Design
Multi-institutional model development and external validation study
Authors
Loading...
AI ECG models for low EF show institution-dependent external validity; leaves open optimal training for broad applicability.
Does an AI-enabled 12-lead ECG model accurately detect ejection fraction <40% across different institutions and patient strata?
AI models for detecting low ejection fraction from ECGs show variable external validity depending on the training dataset, emphasizing the critical need for external validation.
Yagi et al. (2022) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: