Why the study?
CNNs for ECG analysis require large sample sizes, and transfer learning pre-trained on natural images may yield suboptimal performance in biomedical applications.
Does a foundational vision transformer model (HeartBEiT) pre-trained on ECGs improve diagnostic classification performance for LVEF ≤40%, HCM, and STEMI compared to standard CNNs?
Population
8.5 million ECGs
Comparison
Vision transformer model HeartBEiT vs standard CNN architectures
Design
Model development and comparative validation study
Authors
Loading...
Supports ECG-based LVEF screening with minimal data; extends ECG-pretrained transformers beyond CNN benchmarks in cardiology.
Does a foundational vision transformer model (HeartBEiT) pre-trained on ECGs improve diagnostic classification performance for LVEF ≤40%, HCM, and STEMI compared to standard CNNs?
A foundational vision transformer pre-trained on ECGs significantly outperforms standard CNNs trained on natural images for diagnosing cardiac conditions, especially in data-limited settings.
Vaid et al. (2023) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: