Effectiveness of AI-assisted ESI triage on accuracy and selected outcomes in emergency nursing: A systematic review

To evaluate the effectiveness of artificial intelligence (AI) assisted Emergency Severity Index (ESI) triage systems in improving triage accuracy, selected outcomes including under-triage and over-triage, waiting time and patient workflow, and barriers to implementation in emergency nursing.

Systematic review.

A narrative synthesis was used to evaluate findings from eligible studies. The Mixed Methods Appraisal Tool (MMAT) was applied for quality assessment. Studies were included if they examined AI-assisted ESI triage systems involving emergency nurses and reported on triage performance and implementation challenges.

Search was performed in CINAHL, Medline, PsycINFO, PubMed, and Google Scholar for English-language articles published between 2018 and 2025.

Ten studies met the inclusion criteria. AI-assisted ESI triage systems improved accuracy, demonstrating higher AUC, F1 score, sensitivity, and specificity compared to traditional triage nursing. These systems also reduced rates of over-triage and under-triage, minimized long waiting times, and enhanced patient flow. However, barriers included reliance on retrospective data, the need for model validation, and potential resistance from nurses.

AI-assisted ESI triage systems demonstrate promising benefits in enhancing triage accuracy and efficiency in emergency nursing. While AI can be a valuable decision-support tool, it should complement rather than replace clinical judgment. Integrating AI into emergency triage may streamline workflows, reduce workload, and improve the accuracy of patient assessments.