
Researchers at the University of California, Berkeley have developed a deep-learning algorithm that detects a previously unrecognized biological warning sign in routine electrocardiogram (EKG) scans capable of predicting sudden cardiac death before it strikes. Outperforming standard clinical screening methods, the AI model identifies subtle electrical anomalies in patients whose hearts otherwise appear healthy. Beyond enabling early life-saving interventions such as continuous wearable monitoring or implantable defibrillators, the discovery demonstrates how artificial intelligence can uncover hidden physiological patterns that have long eluded human observation.