Title: AI-Powered CDSS for the Interpretation of ST-Segment Elevation Myocardial Infarctions in Electrocardiograms: A Systematic Literature Review
Authors: Michael Etter
Dr. Bruce Lazar
Volume: 10
Issue: 8
Pages: 114-120
Publication Date: 2026/08/28
Abstract:
This systematic review examines the diagnostic accuracy of artificial intelligence (AI)-powered clinical decision support system (CDSS) in the interpretation of ST-segment elevation myocardial infarctions (STEMIs) in electrocardiograms (ECGs) in adult inpatients. Hospitals today continue to struggle to meet the national standards for first-medical-contact to device-times even with strict STEMI criteria. This poor triaging of patients leads to missed device times and poor patient outcomes. Limited evidence on the use of AI-powered CDSS is published. A systematic literature review of current research is required to evaluate the diagnostic accuracy of AI-powered CDSS for interpreting STEMIs on adult inpatients. This review utilized the PRISMA 2020 guidelines to identify studies published between the years 2021-2026 using the databases Medline Ultimate, PubMed, and CINAHL. Keyword searches included ST-segment elevation myocardial infarction (STEMI), AI-assisted ECG interpretation, diagnostic accuracy, and ECG interpretation, which were combined with Boolean operators (AND/OR) and MeSH terms. Other inclusion criteria included peer-reviewed articles, articles published in English, and full-text articles. Studies that met these criteria were then screened by titles/abstract. A full-text review was then conducted for the final inclusion process. Utilizing this screening process, eighteen studies were included in the systematic review that included cohort and diagnostic accuracy studies. Primary outcomes found consistently high diagnostic accuracy for AI-powered CDSS. Secondary outcomes demonstrated improved ECG turnaround times and improved door-to-balloon times with the use of AI-powered CDSS. Evidence gaps included differing AI algorithms utilized throughout the included studies. The studies were also performed across a broad range of patient populations, which influences variability in outcomes. The overall findings demonstrated improved diagnostic accuracy of STEMI identification in ECGs along with improved ECG turnaround times and door-to-balloon times. Due to the differing algorithms, future research needs to focus on a single AI algorithm for ease of implementation into workflows.