Patients with occlusion myocardial infarction (OMI) and no ST-elevation on presenting electrocardiogram (ECG) are increasing in numbers. These patients have a poor prognosis and would benefit from immediate reperfusion therapy, but, currently, there are no accurate tools to identify them during initial triage. Here we report, to our knowledge, the first observational cohort study to develop machine learning models for the ECG diagnosis of OMI. Using 7,313 consecutive patients from multiple clinical sites, we derived and externally validated an intelligent model that outperformed practicing clinicians and other widely used commercial interpretation systems, substantially boosting both precision and sensitivity. Our derived OMI risk score provided enhanced rule-in and rule-out accuracy relevant to routine care, and, when combined with the clinical judgment of trained emergency personnel, it helped correctly reclassify one in three patients with chest pain. ECG features driving our models were validated by clinical experts, providing plausible mechanistic links to myocardial injury.

Background: High-sensitivity cardiac troponin (hs-cTn) laboratory assays are used to rule out myocardial infarction (MI) on presentation, but prolonged result turn-around times can delay patient management. Our primary aim was to identify patients at low risk of index MI using a rapid, point of care (POC), whole blood hs-cTnI assay at presentation, with potential early patient discharge. Methods: Consecutive emergency department patients from two prospective, observational studies with suspected acute coronary syndrome were enrolled. A POC hs-cTnI assay (Atellica^® VTLi) threshold using whole blood at presentation, that resulted in a negative predictive value (NPV) of ≥99.5% and sensitivity of ≥99% for index MI, was derived (Safe Emergency department dIscharGE rate [SEIGE]) and validated using plasma (Suspected acute myocardial infarction in Emergency [SAMIE]). Event adjudications were established using hs-cTnI assay results from routine clinical care. The primary outcome was MI at 30-days. Results: 1086 patients (8.1% MI) were enrolled in a US derivation cohort (SEIGE) and 1486 (5.5% MI) in an Australian validation cohort (SAMIE). A derivation whole blood POC hs-cTnI concentration of <4ng/L provided a sensitivity of 98.9% (93.8-100%) and NPV of 99.5% (95% CI: 97.2-100%) for ruling out MI. In the validation cohort, sensitivity was 98.8% (93.3-100%) and NPV was 99.8% (99.1-100%). 17.8% and 41.8%, respectively, were defined as low risk for discharge. 30-day adverse cardiac events were 0.1% (n=1) for SEIGE and 0.8% (n=5) for SAMIE. Conclusions: A point-of-care, whole blood hs-cTnI assay permits an accessible, rapid and safe exclusion of MI, and may expedite discharge from the emergency department.