(P046) UNVEILING HIDDEN RISKS: ENHANCING DETECTION OF TOTALLY OCCLUDED ARTERIES IN NON-ST-ELEVATION MYOCARDIAL INFARCTION PATIENTS WITH ARTIFICAL INTELLIGENCE

Background: Patients with a Non-ST-Elevation Myocardial Infarction (NSTEMI) and totally occluded arteries (TOA) experience increased morbidity and mortality compared with NSTEMI without TOA. NSTEMI with TOA also undergo delayed revascularization treatments as compared to STEMI. The Occlusion Myocardial Infarction Artificial Intelligence Model (OMI-AI-Model) used within the PMCardio application aims to improve TOA detection, irrespective of a STEMI diagnosis and was twice as sensitive at detecting OMIs than standard care, translating into a quicker diagnosis. The left circumflex (LCx) vessels account for half of TOA presenting with NSTEMI. Our study aims to evaluate OMI AI Model’s ability to detect OMIs at presentation in NSTEMI patients that have a 100% LCx culprit lesion occlusion on invasive angiogram (IA).

METHODS AND RESULTS: Retrospective review included consecutive patients admitted with an NSTEMI at three local hospitals between July 1, 2022 to June 30th, 2023 who had a culprit lesion TOA of the LCx (including Obtuse Marginal branches or vein graft to these vessels) on IA. Data collected from the hospitals’ electronic medical records, local CorHealth cardiac catheterization registry, and OMI-AI-Model included patient demographics, angiogram procedural and laboratory data, and ECG interpretations. The authors were given complimentary access to PMCardio and the OMI- AI-Model, solely for the purposes of this study (REB- 17328).
27 patients were found to be NSTEMIs with a true culprit lesion involving the LCx vessels on chart review (i.e. not a Chronic-Total-Occlusion with a culprit in another vessel). 23(85%) were male, the mean age was 70 years (SD8), 15(56%) had diabetes, 20(74%) had a history of smoking, the mean creatinine was 102.7 µmol/L (SD39), 17(63%) had a prior MI, 12(44%) had prior CABG, 17(63%) had prior PCI, and the mean peak troponin was 18479 µmol/L (SD36737). Applying OMI-AI-Model to all pre-IA ECGs, the application identified an OMI in 17(63%) patients. Mean time from initial ECG to IA was 2102 minutes (SD2998) in these 17 patients. Mean time from initial ECG to diagnostic ECG of OMI identified by OMI-AI-Model was 91 minutes (SD224). Given the mean time from ECG diagnosis of STEMI to IA is 76 min for these three hospitals, use of OMI-AI-Model would reduce the time to IA by a mean of 1937 minutes.

Conclusion: OMI-AI-Model shows promise in identifying OMIs involving the LCx in NSTEMIs, not identified by usual practices, potentially leading to revascularization >24 hours earlier and better clinical outcomes. Further prospective evaluation is recommended.