A NEW era in heart attack treatment is emerging. The GRACE 3.0 scoring system has been validated across ten countries, using AI to more accurately predict outcomes and personalise care for patients with non-ST-elevation acute coronary syndrome (NSTE-ACS).
AI Drives Advances in the GRACE 3.0 Scoring System
The GRACE 3.0 scoring system represents a major upgrade to one of cardiology’s most trusted risk tools. For decades, doctors have relied on the GRACE score to guide treatment decisions for patients with NSTE-ACS, the most common form of heart attack. However, earlier versions could not fully account for individual differences in patient response or long-term mortality. By integrating AI-driven models, the latest update provides a refined method for assessing both short-term and long-term mortality risk, as well as identifying who benefits most from early invasive management, such as angiography and stent placement.
Validation Across Over 600,000 Patients Worldwide
Researchers analysed data from 609,063 patients with NSTE-ACS across 10 countries between 2005 and 2024. The updated GRACE 3.0 scoring system incorporated machine learning to enhance predictive accuracy. Importantly, they re-analysed data from the landmark VERDICT trial, teaching the model to recognise which patients benefit most from early invasive treatment including angiography and stenting. Corresponding models for in-hospital mortality and 1-year mortality were created, in addition to a separate machine learning model developed to predict the individualised effect of early versus delayed invasive coronary angiography and revascularisation at a median follow-up of 4·3 years.
The in-hospital mortality model achieved an area under the receiver operating characteristic curve (AUC) of 0.90 (95% CI 0.89–0.91), while the 1-year mortality model reached an AUC of 0.84 (95% CI 0.82–0.86). Compared with version 2.0, both models showed improved risk reclassification and calibration. Additionally, the individualised treatment effect model identified patients who gained significant benefit from early invasive treatment, with a hazard ratio of 0.60 (95% CI 0.41–0.88) for high-benefit groups, compared with 1.06 (0.80–1.40) for those with little to no expected benefit (pinteraction=0.014). This data-driven precision suggests that current treatment guidelines may be overlooking key subgroups of patients who could benefit most from timely intervention.
Transforming Clinical Practice Through Personalised Risk Prediction
The findings suggest that the GRACE 3.0 scoring system could fundamentally change how cardiologists manage NSTE-ACS. By combining validated AI models with clinical insight, it offers a practical and personalised tool that refines treatment decisions and potentially saves lives. Future guidelines may soon integrate this model, marking a significant step toward precision cardiology on a global scale.
Reference
Wenzl FA et al. Extension of the GRACE score for non-ST-elevation acute coronary syndrome: a development and validation study in ten countries. The Lancet Digital Health. 2025;DOI: 0.1016/j.landig.2025.100907.
