AI-Powered ECG Interpretation Found Better than Standard ECG in Acute MI- Insights From the Queen of Hearts Registry: TCT 2025

The Queen of Hearts registry, a U.S. multicenter study, evaluated the performance of an AI-based ECG algorithm trained to detect angiographically confirmed STEMI and STEMI-equivalent events. The registry included 1,085 consecutive STEMI activations across three primary PCI centers between January 2020 and May 2024. Of these, 1,032 patients were eligible for analysis—601 with true STEMI (58%) and 431 with STEMI mimics (42%).

The model, trained using deep neural networks validated against angiographic confirmation, was developed to enhance STEMI detection accuracy while minimizing unnecessary activations. In the registry, baseline characteristics were well balanced between true STEMI and mimic groups (mean age 63 years; ~70% men). Patients commonly presented via emergency medical services (≈60%), while approximately 20% were walk-ins or transfers.

The AI-powered ECG model demonstrated marked improvement in diagnostic performance. Compared with standard-of-care ECG interpretation, the AI model achieved a sensitivity of 92% versus 71%, specificity of 81% versus 29%, and reduced false-positive activation rates from 41.8% to just 7.9%. The model also identified more true STEMI cases on the initial ECG, particularly those with atypical or “STEMI-equivalent” presentations, such as left circumflex occlusions or evolving infarctions.

Detailed analysis showed that conventional ECG interpretation often misclassified non-ischemic patterns—including left bundle branch block, left ventricular hypertrophy, and nonspecific ST-T changes—as STEMI, leading to unnecessary cath lab activations. The AI model reclassified the majority of these cases correctly as non-STEMI mimics, reducing false activations by fourfold and improving diagnostic consistency across sites.

While clinical outcomes such as infarct size and mortality were not assessed in this analysis, procedural metrics supported the AI model’s potential to streamline workflows. Among confirmed STEMI patients, door-to-balloon times under 90 minutes were achieved in 84.6% of conventional STEMI and 73.5% of STEMI-equivalent cases, with percutaneous coronary intervention (PCI) performed in approximately 85%.

The investigators acknowledged that this registry was retrospective and limited to STEMI activations at high-volume centers, necessitating prospective trials to confirm real-world clinical benefits. 

The Queen of Hearts registry, supported by Powerful Medical Inc., represents one of the first large-scale validations of AI-based ECG interpretation in STEMI care, offering a promising path toward earlier, more accurate myocardial infarction detection and fewer unnecessary cath lab activations in the U.S. health system.

Reference: Timothy D. Henry, AI-Powered ECG Interpretation Enables Earlier STEMI Identification and Reduces False-Positive Activations: A U.S. Multi-center Registry, TCT Conference 2025, San Francisco.

https://www.tctconference.com/

About the Study Presenter: Timothy D. Henry, MD, MSCAI, is the Medical Director of The Carl and Edyth Lindner Center for Research and Education at The Christ Hospital in Cincinnati, Ohio. He has published over 750 manuscripts and book chapters and has served as principal investigator and steering committee member of multiple large, multicenter research trials in acute myocardial infarction, refractory angina, and regenerative medicine with gene and stem cell therapy.

Among other awards, Dr. Henry has been named to the Best Doctors in America list each year for the last 15 years, became a Master Fellow of SCAI in 2015, received the American Heart Association’s Heart and Stroke Hero Award in Research in 2013, the LUMEN Global Lifetime Achievement Award in MI in 2012, and most recently was awarded the Cincinnati 2021 Health Care Heroes Award, Innovator.