Artificial Intelligence–Powered Rapid Identification of ST-Elevation Myocardial Infarction via Electrocardiogram (ARISE) — A Pragmatic Randomized Controlled Trial

BackgroundTimely diagnosis of ST-elevation myocardial infarction (STEMI) is crucial for the treatment of patients with acute coronary syndrome. Artificial intelligence–enabled electrocardiogram (AI-ECG) has shown potential for the accurate and timely detection of STEMI on 12-lead electrocardiograms (ECGs). However, its impact on clinical treatment times is unknown.MethodsTo evaluate the potential of AI-ECG–assisted detection of STEMI to reduce treatment delays for patients with STEMI, we conducted an open-label, cluster randomized controlled trial involving 43,234 eligible patients (mean age, 60 years; 49.5% male) without a history of coronary angiography within 3 days in the emergency department or inpatient wards at Tri-Service General Hospital, Taipei, Taiwan between May 1, 2022, and April 31, 2023. Patients were randomly assigned 1:1 to AI-ECG–assisted detection of STEMI (intervention group) or to standard of care (control group). The primary end point was door-to-balloon time; ECG-to-balloon time was also evaluated as a branch of the primary analysis. Secondary end points included incidence of new-onset low ejection fraction, cardiac death, and all-cause mortality.ResultsAmong the 43,234 patients, 77 in the intervention group and 68 in the control group were diagnosed with STEMI with occluded vessel(s) based on coronary angiography. The use of AI-ECG demonstrated a positive predictive value of 89.5% (95% confidence interval CI, 85.3 to 93.6%) and a negative predictive value of 99.9% (95% CI, 99.9 to 100.0%). For patients in the emergency department, the median door-to-balloon time was 82.0 minutes (interquartile range, 62.5 to 89.5) in the intervention group compared with 96.0 minutes (interquartile range, 78.0 to 137.0) in the control group (P=0.002). When analyzing both emergency and inpatient cases, the median ECG-to-balloon time was 78.0 minutes (interquartile range, 56.9 to 88.2 minutes) in the intervention group compared with 83.6 minutes (interquartile range, 72.7 to 127.8 minutes) in the control group (P=0.011). In the intervention group versus the control group, there were 340 versus 304 cases, respectively, of new-onset heart failure with reduced ejection fraction (odds ratio, 1.12; P=0.151), 85 versus 116 cases of cardiac death (odds ratio, 0.73; P=0.029), and 1153 versus 1127 cases of all-cause mortality (odds ratio, 1.02; P=0.568).ConclusionsIn patients with STEMI, AI-ECG–assisted triage of STEMI decreased the door-to-balloon time for patients presenting to the emergency department and decreased the ECG-to-balloon time for patients in the emergency room and inpatients. (Funded by the National Science and Technology Council, Taiwan and others; ClinicalTrials.gov number, NCT05118009.)