SPC365

PRELIMINARY RESULTS FROM THE EXTERNAL VALIDATION OF AN ARTIFICIAL INTELLIGENCE MODEL FOR OCCLUSION MYOCARDIAL INFARCTION DETECTION

Session: SESSÃO DE COMUNICAÇÕES ORAIS 15 - INTELIGÊNCIA ARTIFICIAL EM CARDIOLOGIA: APROVEITAR O POTENCIAL!
Speaker: Mafalda de Oliveira Griné

Congress: CPC 2025
Topic: E. Coronary Artery Disease, Acute Coronary Syndromes, Acute Cardiac Care
Theme: 13. Acute Coronary Syndromes
Subtheme: 13.3 Acute Coronary Syndromes – Diagnostic Methods
Session Type: Comunicações Orais
FP Number: ---
Authors: Mafalda Griné; Catarina Sena Silva; Henrique Sena Silva; Rita Bertão Ventura; Tomás Carlos; Bernardo Resende; Luísa Rocha; Manuel Oliveira-Santos; Miguel Nobre Menezes; Lino Gonçalves

<p style="text-align:justify"><span style="font-size:medium"><span style="font-family:Aptos,sans-serif"><span style="color:#000000"><strong><span style="font-family:&quot;Times New Roman&quot;,serif">Background:</span></strong><span style="font-family:&quot;Times New Roman&quot;,serif"> Around 15 to 30% of patients presenting without significant ST-segment elevation have an acutely occluded coronary artery. These patients have a worse prognosis, likely related to delayed revascularization. We aimed to test a novel artificial intelligence (AI) model designed to enhance the detection of these cases based on admission 12-lead electrocardiograms (ECGs).</span></span></span></span></p>

<p style="text-align:justify">&nbsp;</p>

<p style="text-align:justify"><span style="font-size:medium"><span style="font-family:Aptos,sans-serif"><span style="color:#000000"><strong><span style="font-family:&quot;Times New Roman&quot;,serif">Methods:</span></strong><span style="font-family:&quot;Times New Roman&quot;,serif"> A total of 658 ECGs from 398 patients admitted to the emergency department with suspected acute coronary syndrome (ACS) were retrospectively analyzed via the OMI AI ECG</span> <span style="background-color:white"><span style="font-family:&quot;Times New Roman&quot;,serif"><span style="color:#222222">Model</span></span></span> <span style="font-family:&quot;Times New Roman&quot;,serif">(Powerful Medical, Slovakia). The primary endpoint was the detection of occlusion myocardial infarction (OMI), defined as angiographic evidence of an acute culprit lesion with either 0-2 TIMI flow and positive troponin or TIMI 3 flow and significant troponin elevation (i.e. high-sensitivity troponin I &ge; 5000 ng/L). The model&rsquo;s performance was compared with the current gold standard.</span></span></span></span></p>

<p style="text-align:justify">&nbsp;</p>

<p style="text-align:justify"><span style="font-size:medium"><span style="font-family:Aptos,sans-serif"><span style="color:#000000"><strong><span style="font-family:&quot;Times New Roman&quot;,serif">Results</span></strong><span style="font-family:&quot;Times New Roman&quot;,serif">: In this initial test set, we identified 147 (36.9%) OMI cases. The OMI AI ECG</span> <span style="background-color:white"><span style="font-family:&quot;Times New Roman&quot;,serif"><span style="color:#222222">Model</span></span></span> <span style="font-family:&quot;Times New Roman&quot;,serif">achieved 72% accuracy (95% confidence interval (CI): 67.4&ndash;76.5), 44.2% sensitivity (95% CI: 37.2&ndash;51.6), 91.9% specificity (95% CI: 88.7-94.8), 79.6% PPV (95% CI: 71.9-86.6), NPV 69.8% (95% CI: 63.9-75.4), and a 0.422 Mathew&rsquo;s correlation coefficient (MCC; 95% CI: 0.341-0.503), whereas the ST-segment elevation myocardial infarction (STEMI) criteria had 66.3% accuracy (95% CI: 61.0-71.3), 28.5% sensitivity (95% CI: 22.3-35.3), 93.2% specificity (95% CI: 90.1-96.1), 75.0% PPV (95% CI: 64.8-84.5), 64.6% NPV (95% CI: 58.6-70.4), and a 0.293 MCC (95% CI: 0.206-0.38)]. Demographic parameters, such as age and sex, did not impact model performance. Notably, within the patient group who underwent coronary angiography within 2 hours of admission, the model&rsquo;s sensitivity increased to 81.2% (CI: 73.1-88.5), reflecting good model performance in acute/active case detection.</span></span></span></span></p>

<p style="text-align:justify">&nbsp;</p>

<p style="text-align:justify"><span style="font-size:medium"><span style="font-family:Aptos,sans-serif"><span style="color:#000000"><strong><span style="font-family:&quot;Times New Roman&quot;,serif">Conclusion:</span></strong><span style="font-family:&quot;Times New Roman&quot;,serif"> In this challenging all-comer suspect ACS cohort, the OMI AI ECG</span> <span style="background-color:white"><span style="font-family:&quot;Times New Roman&quot;,serif"><span style="color:#222222">Model</span></span></span> <span style="font-family:&quot;Times New Roman&quot;,serif">outperformed the STEMI criteria in active OMI detection, with about 1.5 times higher sensitivity, without compromising specificity. This tool may contribute to better patient triage and timely revascularization.</span></span></span></span></p>