Background 

​

Current guidelines suggest to classify patients with acute coronary syndromes  according to STEMI criteria, however, this classification has important limitations.  Large series have documented that up to 15% of patients initially classified as NSTEMI  will show evidence of total coronary occlusion on index angiography. Artificial  intelligence (AI) holds promise in this regard, previous endeavors have shown that AI 

EKG based algorithms can enhance diagnostic accuracy, considering OMI signatures  are spatial in nature, representing a significant variation among patients. 

 

Methods  

This is a prospective, cross-sectional study based on the development of an AI-based  ECG model capable of detecting an Acute Coronary Occlusion Myocardial Infarction  (ACOMI). Single-standard 12-lead electrocardiograms from 366 patients who  presented to the emergency department with an acute coronary syndrome between  2017 to 2023 at the National Institute of Cardiology, Mexico City, Mexico were  contemplated. ECGs were also evaluated by two expert cardiologists independently  and blinded for the clinical outcome, each was presented with the images and asked  to determine whether the patient had an STEMI based on the criteria of the fourth  Universal Definition of Myocardial Infarction and the suspicion of OMI based on the  state-of-the-art criteria. ACOMI was angiographically defined as the presence of one  of the following: total thrombotic occlusion, TIMI thrombus grade 2 or higher + TIMI  grade flow 1 or less, or the presence of a subocclusive lesion. Patients were classified  into four groups: ST elevation + OMI (Occlusion Myocardial Infarction), No ST  elevation + OMI, ST elevation + NOMI (Non-occlusion Myocardial Infarction)and No  ST elevation + NOMI.  

 

Results  

Patients were 62.95 ± 11.57 years old, 80.47% male, and the proportion of diabetes  was 41.71%, hypertension and dyslipidemia were 52.71% and 28.45%, respectively. 

The ACOMI AI model accomplished an AUC of 0.87 in identifying occlusion myocardial  infarction as our primary outcome, compared to the AUC achieved by ECG experts,  AUC: 0.53. Our model showed a higher sensitivity (83-86%) in identifying OMI  compared to the STEMI criteria (70%) and ECG experts (80%).  

 

Conclusion 

This study aimed to assess the performance of an AI-EKG based algorithm capable  of detecting ACOMI in the setting of patients with non ST-segment elevation  myocardial infarction (or NSTEMI). In the present study including patients with ACS,  an AI-EKG based algorithm outperformed EKG experts and demonstrated a higher  diagnostic precision for the detection of ACOMI. Further research is needed in order  to externally validate our algorithm and understand the role of AI-based models in the  care of patients with ACS.