AI Heart Attack Risk Assessment Tool Outperforms Existing Methods

The GRACE score, used globally to guide treatment timing, has been instrumental in patient care but is limited in capturing the full complexity of each case. Now, an international research team has shown that artificial intelligence (AI) can outperform traditional methods, enabling more precise and personalized treatment decisions for millions of patients.

The study, led by the University of Zurich (Zurich, Switzerland), analyzed health data from more than 600,000 patients across ten countries, making it the largest investigation into NSTE-ACS risk modeling to date. The results, published in The Lancet Digital Health, reveal that many patients may need to be reclassified under the new model, potentially transforming global heart attack treatment strategies.

The AI-based system—an enhanced version of the existing GRACE model, called GRACE 3.0—was trained using clinical data from the landmark VERDICT trial to identify which patients benefit most from early invasive procedures like angiography and stenting. By learning directly from clinical trial outcomes, GRACE 3.0 can distinguish between patients who respond positively to early intervention and those who show little or no benefit.

The findings suggest that current treatment strategies may often be misdirected, targeting patients who do not derive significant advantage from invasive procedures while overlooking others who would. This represents a potential paradigm shift toward individualized care guided by AI-driven insights rather than fixed clinical rules.

In addition to refining patient stratification, GRACE 3.0 offers improved accuracy in predicting complications and survival, outperforming traditional methods in all tested datasets. The new model provides a practical, easy-to-use decision support tool that could soon be integrated into hospital workflows, supporting clinicians in tailoring therapy timing and intensity to each patient’s unique profile.

“GRACE 3.0 is the most advanced and practical tool yet for treating patients with the most common type of heart attacks,” said last author, Thomas F. Lüscher. “This could reshape future clinical guidelines and help to save lives.”

Related Links:
University of Zurich