What is AI-assisted OMI/NOMI paradigm in ACS?

The AI-assisted OMI/NOMI paradigm represents a fundamental shift in the classification and acute management of acute coronary syndrome (ACS). It replaces the traditional, millimeter-based STEMI/NSTEMI dichotomy with a framework focused entirely on the underlying pathophysiology—specifically, the presence or absence of an acute coronary occlusion—and utilizes deep-learning artificial intelligence to detect the subtle occlusions that standard criteria miss.

The Paradigm Shift: STEMI/NSTEMI to OMI/NOMI

The traditional STEMI criteria are highly specific but poorly sensitive. Angiographic data consistently shows that approximately 25% to 30% of patients classified clinically as having an NSTEMI actually have an acute coronary occlusion with TIMI 0-2 flow and lack sufficient collateral circulation.

• Occlusive Myocardial Infarction (OMI): Patients with an acute or near-acute coronary occlusion requiring emergent reperfusion therapy (PCI or thrombolysis), regardless of whether their ECG meets standard ST-elevation millimeter criteria. This encompasses both STEMI(+) OMI and STEMI(-) OMI.
• Non-Occlusive Myocardial Infarction (NOMI): Patients without an acute occlusion or those with sufficient collateral circulation such that they do not require emergent catheterization to salvage myocardium.

When STEMI(-) OMI patients are treated under the standard NSTEMI pathway, reperfusion is often delayed by 24 to 72 hours, resulting in significantly higher morbidity and nearly double the mortality compared to true NOMI cases.

The Role of AI in Detection

While experienced clinicians can recognize the subtle, “non-STEMI” ECG signatures of an acute occlusion, these morphologies are notoriously difficult to standardize and teach broadly across rapid triage environments.

Deep neural network models bridge this gap by training on hundreds of thousands of ECGs linked directly to angiographic outcomes rather than human-annotated interpretations. Because the AI is optimized against the actual catheterization results (the ground truth), it accurately identifies complex, non-linear relationships and markers of STEMI(-) OMI, including:

• Hyperacute T-waves: Disproportionate T-wave amplitude or volume relative to the QRS complex.
• Terminal QRS distortion: The absence of both S-waves and J-waves in leads V2-V3.
• Subtle reciprocal changes: Specifically ST depression in lead aVL, which can be the earliest or only sign of an inferior OMI.
• High-risk equivalents: de Winter patterns, Aslanger’s pattern, and subtle posterior OMIs.

By analyzing these features simultaneously, AI-assisted triage models routinely achieve a sensitivity of roughly 90% for detecting true OMI—significantly outperforming the ~60-70% sensitivity of standard STEMI criteria—while simultaneously reducing false-positive cath lab activations.

References

Kola M, Shuka N, Meyers HP, Zaimi Petrela E, Smith SW. OMI/NOMI: Time for a New Classification of Acute Myocardial Infarction. J Clin Med. 2024 Sep 2;13(17):5201. doi: 10.3390/jcm13175201. PMID: 39274412; PMCID: PMC11395726.

Aslanger EK, Aggül B, Yıldırımtürk Ö, Karabay CY, Meyers HP, Smith SW, Değertekin M; DIFOCCULT-3 Study Investigators. A Diagnostic Paradigm Shift in Acute Myocardial Infarction: Rationale and Design of the DIFOCCULT-3 Trial. JACC Adv. 2025 Nov;4(11 Pt 2):102227. doi: 10.1016/j.jacadv.2025.102227. Epub 2025 Oct 22. PMID: 41128712; PMCID: PMC12717606.