Convolutional and transformer-based neural networks trained on millions of ECG recordings now match or exceed cardiologist accuracy for arrhythmia detection and STEMI identification. In the emergency setting, these tools are dramatically compressing the time from ECG acquisition to treatment initiation.