{
  "abstract": "Introduction Rapid and accurate electrocardiogram (ECG) interpretation is critical for early diagnosis of life-threatening cardiovascular conditions in emergency care, yet clinicians frequently work under intense time pressure with substantial inter-observer variability. Artificial intelligence (AI)-assisted ECG analysis has emerged as a digital innovation with potential to augment diagnostic accuracy, optimise triage, and improve time-critical care pathways. This systematic review evaluates the diagnostic performance and clinical utility of AI-assisted ECG interpretation compared with standard clinician interpretation in acute and emergency care, with a focus on implications for cardiovascular service delivery.Methods A systematic review was conducted in line with PRISMA guidance. PubMed and the Cochrane Library were searched for the last 10 years for studies of adult patients (≥18 years) in emergency or acute care settings that compared AI-assisted ECG interpretation with clinician interpretation. Eligible studies reported diagnostic accuracy and/or clinical workflow outcomes. Data were extracted on study design, setting, AI model architecture, target condition, reference standard, and reported outcomes. Risk of bias was assessed using QUADAS-2 for diagnostic accuracy studies and ROBINS-I for observational studies. Owing to heterogeneity in AI models and outcome reporting, findings were synthesised narratively.Results Twelve studies including over 100,000 adults met the inclusion criteria. AI-assisted ECG interpretation demonstrated high diagnostic accuracy overall, with a pooled area under the curve of 0.92. Across several high-risk conditions, AI outperformed clinicians: for occlusion myocardial infarction, AUC 0.91 vs 0.79; for hyperkalaemia, AUC 0.90 vs 0.66 (p < 0.001). In emergency triage pathways, AI-enabled models safely ruled out approximately 55% of chest pain presentations for 30-day major adverse cardiac events, with negative predictive value exceeding 99.5%. Integration of AI-assisted ECG was associated with an 85% improvement in occlusion myocardial infarction detection and an average 11-minute reduction in door-to-balloon time for ST-elevation myocardial infarction. Most studies were single-centre and retrospective, with variable implementation strategies and moderate heterogeneity.Conclusions This systematic review indicates that AI-assisted ECG interpretation consistently improves diagnostic accuracy and can deliver clinically meaningful enhancements in triage performance and emergency department workflow. AI-ECG therefore represents a promising digital service innovation and decision-support tool, complementing rather than replacing clinician expertise. Prospective multicentre implementation studies focused on patient-centred outcomes are now required to define how AI-ECG can be safely and effectively integrated into UK cardiology services, emergency care pathways, and training programmes.",
  "authors": [
    {
      "affiliations": [
        "University of Buckingham, Buckingham, United Kingdom"
      ],
      "name": "Alishba Awais"
    },
    {
      "affiliations": [
        "University of Buckingham, Buckingham, United Kingdom"
      ],
      "name": "Karan Raj Nazareth"
    }
  ],
  "title": "565 AI-assisted electrocardiogram interpretation in emergency care: systematic review of diagnostic accuracy and clinical utility",
  "uid": "34e75d62-ac42-5697-ae4c-c24f8cb4862a"
}