AI may spot deadly heart risk in a routine ECG

UC Berkeley researchers developed an AI model trained on over 440,000 ECGs to identify patterns linked to sudden cardiac death, outperforming traditional methods like left ventricular ejection fraction (LVEF). The model detected a high-risk group with a 7% annual sudden cardiac death rate, compared to 4.6% for the standard LVEF method, highlighting missed risks in routine ECGs.