Researchers report that an artificial-intelligence model can identify people at elevated risk of sudden cardiac death using routine electrocardiogram (ECG) recordings. The work, described as a new study published in Nature and led by researchers at UC Berkeley, trains the AI on large amounts of medical data, including ECGs and death records. Across sources, the approach is characterized as finding a warning sign in the heart’s electrical activity that is not apparent through conventional ECG screening. In particular, the findings are reported to apply even to some individuals who would otherwise appear healthy under standard assessments. The AI is presented as a tool that could support earlier identification of high-risk patients, enabling clinicians to monitor those individuals more closely. Multiple outlets frame the technology as augmenting clinical judgment rather than replacing doctors. The studies also emphasize that the hidden signal is newly uncovered through data-driven pattern recognition. Overall, the reporting focuses on the potential for earlier risk stratification from routinely collected ECG data, pending further validation and clinical implementation.