Researchers at the University of Pennsylvania have developed an AI model, CAMEL, that can forecast cardiac arrest 10–15 minutes before it occurs by analyzing extended ECG data as if it were language.