U.S. researchers develop AI model to improve sudden cardiac death prediction

Researchers at Johns Hopkins University have developed a new artificial intelligence model that significantly outperforms current clinical guidelines in identifying patients at high risk of sudden cardiac death, according to a newly published study.

The AI system, known as Multimodal AI for Ventricular Arrhythmia Risk Stratification (MAARS), integrates cardiac MRI images with a wide range of patient health records to detect hidden warning signs – offering a new level of precision in cardiovascular risk prediction.

The study, published this week in Nature Cardiovascular Research, focused on hypertrophic cardiomyopathy – one of the most common inherited heart conditions and a leading cause of sudden cardiac death in young people.

"Currently, we have patients dying in the prime of their life because they aren't protected, and others who are putting up with defibrillators for the rest of their lives with no benefit," said senior author Natalia Trayanova, a researcher focused on using AI in cardiology. "We have the ability to predict with very high accuracy whether a patient is at very high risk for sudden cardiac death or not."

Clinical guidelines used in the U.S. and Europe currently have an estimated accuracy of only 50 percent in identifying at-risk patients. In contrast, the MAARS model demonstrated an overall accuracy of 89 percent, with 93 percent accuracy for patients aged 40 to 60 – the group at greatest risk.

The AI model analyzes contrast-enhanced MRI scans for patterns of heart scarring – something physicians have traditionally found difficult to interpret. By applying deep learning to this underused data, the model identifies key predictors of sudden cardiac death.

"Our study demonstrates that the AI model significantly enhances our ability to predict those at highest risk compared to our current algorithms, and thus has the power to transform clinical care," said co-author Jonathan Chrispin, a Johns Hopkins cardiologist.

The team plans to further test the model on a broader patient population and expand the algorithm's use to other heart diseases, including cardiac sarcoidosis and arrhythmogenic right ventricular cardiomyopathy.