National Heart Centre Singapore AI Platform Reads Heart Attack Damage In Under A Minute

Singapore, 25 June 2026 – The National Heart Centre Singapore (NHCS) has developed an artificial intelligence (AI) platform that can analyse the extent of heart muscle damage following a heart attack in under 60 seconds. By comparison, manual analysis by experts can take up to an hour. The platform, called CARDIA-GM, achieves 95% accuracy, and the research validating it has been accepted by the Journal of Cardiovascular Magnetic Resonance, one of the field's leading peer-reviewed journals¹. 

The platform works by automatically analysing cardiac magnetic resonance imaging (MRI) scans to detect and measure two critical markers of post-heart attack damage: scarred heart muscle tissue and microvascular obstruction (MVO), which refers to blockages in tiny blood vessels that impede recovery. Patients with more than 20% heart muscle scarring face approximately three times the risk of a future cardiac event, while those with significant MVO face up to six times higher risk. Identifying these patients rapidly and accurately is essential to tailoring treatment and protecting long-term outcomes.

WHY SPEED AND PRECISION MATTER 

Cardiovascular disease is the leading cause of death in Singapore, accounting for nearly one-third of all fatalities², with more than 13,000 heart attack episodes recorded annually³. After a heart attack, the extent of damage to the heart muscle is one of the strongest predictors of long-term prognosis. Yet the current standard assessment, which relies on expert visual review of cardiac MRI images, is time-intensive and can take up to an hour, with results that vary between clinicians. This has made detailed cardiac MRI damage assessment difficult to scale across routine clinical settings, even where expertise exists, and limits the ability to track changes over the course of a patient's condition. 

CARDIA-GM addresses these limitations. The platform uses advanced machine learning to automatically detect and measure both scarred tissue and MVO, delivering three-dimensional visualisation and measurements within 30 to 60 seconds, independent of the reviewing clinician. Its performance was validated against expert annotations across approximately 3,000 images from over 500 patients at three healthcare centres, including NHCS, and with an independent cohort followed up for up to eight years, demonstrating consistent accuracy across different hospitals, scanner types and patient populations.

FROM ANALYSIS TO PROGNOSIS 

Beyond speeding up analysis, CARDIA-GM shows early promise in helping to identify which patients face the greatest risk of future complications. In the validation study, patients with AI-derived scarring above a defined threshold had significantly lower event-free survival over the follow-up period, with results that were consistent with those from expert manual review. These early findings suggest the platform may offer prognostic value comparable to expert assessment, supporting its potential for clinical risk stratification after a heart attack. 

To illustrate how this works in practice: consider a 72-year-old woman with high blood pressure and high cholesterol admitted after a major heart attack. Following emergency intervention to restore blood flow, she undergoes a cardiac MRI four days later. With CARDIA-GM, the full three-dimensional analysis of her cardiac MRI is completed in under a minute, accurately quantifying the extent of damage to the front wall of her heart, and detecting blockages in tiny vessels that are difficult to identify through manual analysis. The findings reveal her heart is pumping at only 34% of normal capacity, allowing her care team to immediately personalise her treatment strategy, rather than waiting for a lengthy manual review. 

"Cardiac MRI is increasingly being used after heart attacks to assess the extent of damage and guide treatment. CARDIA-GM allows clinicians to rapidly identify patients at highest risk for future cardiac events,” said Adjunct Clinical Associate Professor Tan Ru San, Senior Consultant, Department of Cardiology, NHCS. “When we can determine quickly that a patient has significant heart muscle damage, we can immediately put in place the appropriate treatment strategy for that patient. What previously required an hour of expert analysis can now be done in under a minute, with the same clinical reliability."

"CARDIA-GM reflects what we are working towards in the CVS.AI laboratory: translating rigorous research into practical tools that clinicians can rely on to continue to enhance clinical decision-making," said Associate Professor Zhong Liang, Co-Director and Core Technical Lead of CardioVascular Systems Imaging and Artificial Intelligence (CVS.AI) Research Laboratory, and Principal Investigator and Senior Clinician-Innovator, National Heart Research Institute Singapore, NHCS. "This external validation across multiple centres and scanners gives us confidence that CARDIA-GM can perform consistently in real-world clinical environments, not just in controlled research settings.”

PATHWAY TO CLINICAL DEPLOYMENT

CARDIA-GM has received funding support from the National Health Innovation Centre (NHIC). The platform is currently in a 12-month development phase, with plans for deployment across multiple medical centres in Singapore for validation in the wider population. Beyond its direct clinical application, CARDIA-GM may also support the development of new cardiac therapies. By providing standardised, objective measurements of heart damage, it offers a more precise tool for assessing clinical trial outcomes, potentially shortening the time needed to evaluate whether a treatment is working. 

"NHCS is committed to bringing AI innovations into clinical practice in ways that make a tangible difference for patients,” said Professor Yeo Khung Keong, Chief Executive Officer, NHCS. “CARDIA-GM is a strong example of how our research teams are translating scientific excellence into clinical tools that directly benefit patients. Our aim is to ensure every patient who has had a heart attack benefits from the most precise, timely assessment available, as we continue to raise the standards of heart care for our population."

References:

1.     Yang, P., Leng, S., Zong, D., Hu, M., Tan, R. S., Xiao, X., Sia, C. H., Teo, L., Leiner, T., Ong, C. C., Koh, A. S., Tan, S. Y., Gong, L., Hausenloy, D. J., Chan, M. Y., & Zhong, L. (2026). Prognostic value of end-to-end deep learning assessment of myocardial scar and microvascular obstruction on late gadolinium enhancement cardiovascular magnetic resonance. Journal of cardiovascular magnetic resonance: official journal of the Society for Cardiovascular Magnetic Resonance, 28(1), 102712. https://doi.org/10.1016/j.jocmr.2026.102712

2.     https://www.moh.gov.sg/resources-statistics/singapore-health-facts/principal-causes-of-death

3.     Singapore Myocardial Infarction Registry Annual Report 2022. National Registry of Diseases Office, December 2024.