AI spots heart scarring behind sudden cardiac death
More than 350,000 people in the US die suddenly from cardiac arrest each year. Many appear healthy beforehand. A new AI model has found a pattern that clinicians have consistently missed.
Sudden cardiac death is one of medicine’s oldest mysteries. An implantable defibrillator can prevent it, but only if you know who is at risk. That is the problem: many victims seem perfectly healthy in the lead-up.
New research published in Nature used artificial intelligence to identify people at highest risk. The model found that a specific pattern of cardiac fibrosis (scattered patches of scar tissue distributed through the heart muscle) was consistently present in the highest-risk patients. Fibrosis has long been recognised as a risk factor, but was underestimated because it is difficult to detect on standard imaging.
What the AI model does differently
The researchers trained their model on heart scans and medical records from people who later experienced sudden cardiac arrest. It learned to recognise patterns that the human eye misses. Cardiac fibrosis emerged as a consistent feature in those with the highest risk, and the model could detect it before symptoms appeared.
Relevance for ageing
Fibrosis in the heart muscle accumulates with age. It is one of the mechanisms by which the heart becomes stiffer and more prone to dangerous arrhythmias as we get older. If AI can detect this process earlier, it opens a window for preventive treatment when damage is still limited. The results are preliminary and will need to be confirmed in larger, independent cohorts.