AI model flags hidden heart scarring as a marker of sudden cardiac death risk
Researchers at the University of California, Berkeley reported in Nature that an AI model can help identify people at the highest risk of sudden cardiac arrest by detecting cardiac fibrosis — scar tissue scattered through the heart that clinicians had largely treated as benign. Sudden cardiac arrest kills more than 350,000 Americans each year and can often be prevented with an implantable defibrillator; the hard part has been deciding which patients need one.
The finding reframes diffuse fibrosis as a meaningful warning sign rather than incidental damage, offering a more concrete way to flag apparently healthy people who are nonetheless vulnerable. If the model holds up in further validation, it could sharpen the decision of who receives a defibrillator — sparing low-risk patients an invasive device while catching high-risk patients who currently slip through. As with most single-study results, the practical value will depend on testing in larger and more diverse patient populations before it changes clinical practice.