Machine Learning Finds a Previously Unknown ECG Signal Linked to Cardiac Death Risk

Scientific American reports that machine learning identified a previously unrecognized signal in electrocardiograms associated with the risk of cardiac death. The finding could improve clinical risk stratification, but the available item does not provide the model name, cohort size, validation results, effect size, or evidence of causality. Those details are essential for judging whether the signal generalizes across hospitals, devices, populations, and clinical settings, and whether it adds value beyond established ECG and clinical predictors.