Overview
Discordance in ECG interpretation between Emergency Medicine and Cardiology teams is common, and within canceled STEMI activations, a true acute coronary occlusion myocardial infarction (OMI) can go unrecognized. This retrospective study examined whether an AI ECG algorithm (Queen of Hearts™) could improve OMI detection in this difficult subset. Across three referral centers, the investigators analyzed 185 activations canceled for not meeting STEMI criteria, of which 17 met the definition of a missed OMI. The AI algorithm identified 16 of 17 cases, far exceeding STEMI criteria in sensitivity (94.1% vs. 47.1%), supporting its use as an adjunct to clinical judgment in ambiguous cases.
Published in: Journal of Electrocardiology
Published on: 01 June 2026
Background
ST-segment elevation myocardial infarction (STEMI) and its equivalents describe the electrocardiogram (ECG) findings of acute coronary occlusion myocardial infarction (OMI). Discordance in ECG interpretation between Emergency Medicine and Cardiology teams is common.
Methods
We conducted a retrospective review of STEMI activations over 17 months. We included cases that were canceled with the rationale of “ECG not meeting STEMI criteria.” We excluded sustained activations, cancellations with alternative rationales, and incomplete records. OMI was defined as an angiographic culprit lesion with TIMI 0 or 1 flow. ECGs were reviewed by the AI algorithm and assessed for STEMI criteria.
Results
Of 1224 STEMI activations, 185 cancellations (15.1%) were included, with 17 patients meeting the study definition of OMI. STEMI criteria demonstrated lower sensitivity for OMI as compared to the AI algorithm (47.1% vs 94.1%, p = 0.005), and a non-significantly lower specificity (66.1% vs 73.2%, p = 0.090). The AI algorithm also demonstrated higher positive and negative likelihood ratios for OMI identification (3.51 and 0.08, respectively) than STEMI criteria (1.39 and 0.80, respectively).
Conclusion
Our data suggests that the AI algorithm may serve as a clinical adjunct to improve interrater reliability between Emergency Medicine and Cardiology teams in OMI identification. Further prospective studies may help evaluate its utility in clinical practice.
Authors: Brandon S. Friedman, Rosa Malloy-Post, Stephen W. Smith, H. Pendell Meyers