Fig. 4 | Scientific Reports

Fig. 4

From: Integrating snapshot ensemble learning into masked autoencoders for efficient self-supervised pretraining in medical imaging

Fig. 4

Comparison of performance metrics across models for cardiovascular disease classification using ECG Lead II. Bar plots display the performance of six models (ResNet-34, ViT-S, MAE, and Snap-MAE) across six evaluation metrics: AUC, AUPRC, accuracy, sensitivity, precision, and F1-score. The proposed Snap-MAE (Scalo) consistently achieves high scores, demonstrating superior performance compared to other baselines.

Back to article page