AI-Enabled Decision Support Systems for Patient Triage: A Scoping Review

Background and Rationale: Nurse-led triage in Emergency Departments (ED) is a high-stakes gatekeeping process where rapid prioritization is essential to patient safety. However, the process remains prone to human variability and mistriage. While AI-enabled Clinical Decision Support Systems (CDSS) have emerged as a potential solution to standardize and enhance this process, there is a lack of synthesized evidence regarding how these systems integrate into real-world nursing workflows and their tangible impact on clinical outcomes. Purpose and Objectives: The primary purpose of this scoping review is to map the rapidly evolving landscape of AI and Machine Learning (ML) applications within CDSS specifically designed for nurse-led triage. The review seeks to: Identify the technological architectures (Traditional ML vs. Generative AI) currently in use. Evaluate model performance and the effectiveness of multimodal data fusion (text, vitals, imaging). Identify the socio-technical barriers to implementation, including clinician trust and the "Explainability Paradox." Uncover research gaps concerning the quantification of "soft skills" and non-verbal cues in AI models. Methodology: This review follows the PRISMA-ScR (Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews) framework. A systematic search was conducted across six major databases - Embase, MEDLINE, CINAHL, PubMed, Scopus, and PsycINFO - supplemented by snowballing techniques for the period of 2018–2025.