AI Model Outperforms Doctors in Predicting Sudden Cardiac Arrest Risk

A new artificial intelligence (AI) model has shown significant improvement in predicting which patients are at high risk of sudden cardiac arrest, surpassing the accuracy of traditional clinical methods. The research, led by Johns Hopkins University and funded by the federal government, focuses on hypertrophic cardiomyopathy, a genetic heart condition that can lead to sudden cardiac death, particularly in young individuals and athletes.

Enhancing risk prediction with medical imaging

The AI model, called Multimodal AI for Ventricular Arrhythmia Risk Stratification (MAARS), analyzes a variety of medical records and heart imaging to assess a patient's risk of sudden cardiac death. While the current clinical guidelines used by doctors have only a 50% accuracy rate in identifying patients at high risk, MAARS outperforms these guidelines, with an accuracy rate of 89% across all patients and up to 93% for patients aged 40 to 60, who are at the highest risk.

The breakthrough comes from the model's ability to examine contrast-enhanced MRI images of the heart, a technique that has not previously been used in such detail. The AI identifies patterns of fibrosis—scarring in the heart—that are associated with higher risk but often overlooked in traditional assessments. This capability is crucial because fibrosis is a key marker for sudden cardiac death in patients with hypertrophic cardiomyopathy, yet it has been challenging for doctors to interpret the raw MRI images.

Potential for life-saving interventions

This AI-driven model could help doctors better target their interventions. By predicting which patients are at the greatest risk, it could potentially save lives by recommending preventive measures, such as the implantation of a defibrillator. On the other hand, it could also prevent unnecessary treatments for patients who do not require such interventions.

The research team's model offers the added benefit of explaining why a particular patient is considered high-risk, allowing doctors to develop a personalized treatment plan based on individual needs. This could transform clinical care by improving the precision and effectiveness of medical interventions.

Expanding the model's capabilities

The Johns Hopkins team plans to continue testing the AI model on additional patient groups and hopes to expand its application to other heart conditions such as cardiac sarcoidosis and arrhythmogenic right ventricular cardiomyopathy. Their previous work has also demonstrated the potential of AI to assess survival predictions for patients with heart infarcts, further showing the promise of these models in cardiology.

The study's authors include Changxin Lai, Minglang Yin, Eugene G. Kholmovski, Dan M. Popescu, Edem Binka, Stefan L. Zimmerman, Allison G. Hays from Johns Hopkins, as well as Dai-Yin Lu and M. Roselle Abraham from the Hypertrophic Cardiomyopathy Center of Excellence at the University of California San Francisco, and Erica Scherer and Dermot M. Phelan from Atrium Health. The research was supported by National Institutes of Health grants and a Leducq Foundation grant.

Reference: Rogers AJ, Reynbakh O, Ahmed A, et al. Cardiovascular imaging techniques for electrophysiologists. Nat Cardiovasc Res. 2025. doi:10.1038/s44161-025-00648-8

This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source. Our press release publishing policy can be accessed here.

This content includes text that has been generated with the assistance of AI. Technology Networks' AI policy can be found here.