AI algorithm identifies patients at risk of sudden cardiac arrest

An AI algorithm used with MRI data can predict which patients are at risk of sudden cardiac arrest, researchers have reported.

By analyzing heart imaging results, specifically cardiac MRI, electronic health records, and echocardiograms, the AI algorithm was able to “reveal previously hidden information about a patient’s heart health,” according to a statement released by Johns Hopkins University in Baltimore, at which a team led by Changxin Lai, PhD, conducted the study.

The findings could not only save lives but also avoid unnecessary medical interventions such as the implantation of defibrillators, said senior author Natalia Trayanova, PhD, in the university statement. The work was published on July 2 in Nature Cardiovascular Research.

“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,” Trayanova said. “We have the ability to predict with very high accuracy whether a patient is at high risk for sudden cardiac death or not.”

Hypertrophic cardiomyopathy is one of the most common inherited heart diseases, affecting one in every 200 to 500 individuals worldwide, and is a leading cause of sudden cardiac death in young people and athletes, Lai and colleagues noted. Many people with the condition live normal lives, but some are at increased risk for sudden cardiac death — and it’s difficult for doctors to identify these patients.

Clinical guidelines to find patients most at risk for fatal heart attacks have about a 50% chance of identifying the right ones — “not much better than throwing dice,” Trayanova said. In light of this statistic, the group developed a transformer-based, neural network model called Multimodal AI for ventricular Arrhythmia Risk Stratification (MAARS), using it in a development and validation cohort of 553 patients and another, external cohort of 284 patients. All patients were assessed via traditional clinical guidelines and MR imaging at Johns Hopkins Hospital and Sanger Heart & Vascular Institute in North Carolina.  

Li and colleagues found that MAARS “significantly outperformed” clinical guidelines across all demographics, showing 89% accuracy for predicting sudden cardiac death across all patients and 93% accuracy for people 40 to 60 years old, which is the population among hypertrophic cardiomyopathy patients most at risk.

MARRS’ performance compared to other cardiac death risk assessment tools (internal cross-validation)
Measure ACC and AHA guidelines ESC guideline HCM Risk-SCD Calculator EHR Cardiac imaging report LGE-CMR findings MARRS
Sensitivity 89% 95% 63% 84% 84% 89% 79%
Specificity 31% 15% 47% 72% 62% 75% 82%
Accuracy 54% 50% 55% 77% 72% 81% 80%
AUROC 0.62 0.54 0.54 0.84 0.8 0.86 0.89
ACC = American College of Cardiology; AHA = American Heart Association; AUROC = Area under the receiver operating curve; ESC = European Society of Cardiology; EHR = Electronic health record; LGE-CMR = Late gadolinium enhancement cardiac MRI

“MAARS has the potential to substantially improve clinical decision-making and healthcare delivery for patients with [hypertrophic cardiomyopathy], either directly through future integration with automated data extraction systems or indirectly by serving as a valuable proof of concept for the power of multimodal AI in enhancing personalized patient care,” the investigators wrote.

Going forward, the team plans to test the new model on more patients and expand the algorithm for use with other types of heart diseases, such as cardiac sarcoidosis and arrhythmogenic right ventricular cardiomyopathy, it said.

The complete study can be found here.

Continue Reading