Researchers at the University of California, Berkeley report discovering an electrocardiogram biomarker for sudden cardiac death using a deep learning approach, published in Nature in July 2026, with an Epub date of June 24, 2026. The model analyzes ECGs and identifies a high-risk group, demonstrating a 7.0% annual rate of sudden cardiac death, higher than the 1.9% of the sample with reduced LVEF, which had a 4.6% annual rate. The team found that 86.1% of patients identified as high-risk by the model were not flagged by LVEF, suggesting a potential to improve risk assessment. This new approach, linking ECGs to death certificates, also showed a 54.4% reduction in mortality among high-risk patients who received defibrillators.
Deep Learning Identifies Novel ECG Biomarker for Sudden Cardiac Death
A deep learning model has identified a previously unrecognized biomarker within standard electrocardiograms (ECGs) that improves prediction of sudden cardiac death. This approach moves beyond reliance on left ventricular ejection fraction (LVEF), the current standard, which the team notes misses most sudden cardiac deaths and frequently leads to unnecessary defibrillator implants. The study, published in Nature in July 2026, with an Epub date of June 24, 2026, revealed that the deep learning model identified a high-risk group comprising 2.2% of the analyzed population, experiencing an annual sudden cardiac death rate of 7.0%. This contrasts with a 4.6% annual rate observed in patients with reduced LVEF, demonstrating an improvement in identifying those most vulnerable. The model pinpointed individuals at risk that LVEF overlooked, with research indicating that “86.1% of the model’s high-risk patients were not flagged by LVEF.”
Ziad Obermeyer of the University of California, Berkeley (zobermeyer@berkeley. edu) led the research, utilizing data from the School of Public Health and the College of Computing, Data Science, and Society. Further validation in both a US health system and a Taiwanese hospital registry confirmed the model’s ability to predict life-threatening ventricular arrhythmias and cardiac arrests, respectively. To visualize the waveform morphology discovered by the predictive model, the researchers paired it with a generative model of the ECG waveform. Together, they reveal a biomarker that is easily visible and robustly predicts sudden cardiac death, but has not, to their knowledge, been previously described, prompting the researchers to formulate a new hypothesis regarding the underlying mechanisms of sudden cardiac death. The team’s work suggests that patients receiving defibrillators after identification by the model experienced a 54.4% reduction in mortality compared to expected rates, indicating a potential clinical benefit and a pathway toward more targeted preventative care.
The resulting model isolates a high-risk group (2.2% of the sample) with a 7.0% annual rate of sudden cardiac death, higher than those with reduced LVEF (1.9% of the sample; 4.6% annual rate).
edu), connected comprehensive ECG records from a Swedish region with corresponding death certificates, enabling the development of a predictive model. The study, published in Nature in July 2026, revealed that the deep learning model identified a high-risk group comprising 2.2% of the analyzed population. This contrasts with a 4.6% annual rate observed in patients with reduced LVEF. Notably, 86.1% of the model’s high-risk patients were not flagged by LVEF. Analysis of patients receiving defibrillators based on the model’s predictions revealed a 54.4% reduction in expected mortality, hinting at a potential clinical benefit from targeted intervention. To visualize the waveform morphology discovered by the predictive model, researchers paired it with a generative model of the ECG waveform. Together, they reveal a biomarker that is easily visible and robustly predicts sudden cardiac death, but has not, to our knowledge, been previously described.
