A hidden predictor of sudden cardiac death uncovered by deep learning

NEWS AND VIEWS

24 June 2026 A hidden predictor of sudden cardiac death uncovered by deep learning A machine-learning model trained on thousands of electrocardiogram recordings identifies a previously unrecognized group of at-risk people. By Changxin Lai ORCID: http://orcid.org/0000-0002-3585-5979 0 Changxin Lai Changxin Lai is at Johns Hopkins University, Baltimore, Maryland, 21218, USA, and EnChannel Medical Ltd, Irvine, California, USA. View author publications PubMed Google Scholar

Sudden cardiac death claims hundreds of thousands of lives annually, often striking without warning in people who had seemed reasonably healthy. Implantable defibrillators can terminate the lethal heart rhythms that are responsible, but deciding who should receive a defibrillator depends on accurate risk prediction. The current clinical tools for this miss most people who eventually succumb and flag many who never benefit. Writing in Nature, Obermeyer et al.1 describe a deep-learning model trained on population-scale electrocardiogram (ECG) data and death records. With this model, the authors identify a new high-risk group and discover features of the ECG trace that could be used to predict risk of sudden death.

doi: https://doi.org/10.1038/d41586-026-01806-z

References Obermeyer, Z., Schubert, A., Ross, J., Mullainathan, S. & Lingman, M. Nature https://doi.org/10.1038/s41586-026-10674-6 (2026). Stecker, E. C. et al. J. Am. Coll. Cardiol. 47, 1161–1166 (2006). Merchant, F. M., Quest, T., Leon, A. R. & El-Chami, M. F. J. Am. Coll. Cardiol. 67, 435–444 (2016). Trayanova, N. A. & Topol, E. J. Lancet 399, 1933 (2022). Niederer, S. A., Lumens, J. & Trayanova, N. A. Nature Rev. Cardiol. 16, 100–111 (2019). Download references

Competing Interests The author declares no competing interests.

Subjects