Artificial Intelligence in emergency department triage: A scoping review

Area of Science:

• Medical Informatics
• Clinical Decision Support Systems
• Artificial Intelligence in Healthcare

Background:

• Emergency department (ED) triage is crucial for patient safety and efficient care delivery.
• Existing triage systems face challenges impacting healthcare quality and patient flow.
• Artificial Intelligence (AI) offers potential solutions for optimizing ED decision-making and patient management.

Purpose of the Study:

• To systematically review and map the existing evidence on the implementation and performance of AI in emergency department triage.
• To identify patterns, advances, gaps, and provide recommendations for AI in ED triage.

Main Methods:

• A comprehensive scoping review adhering to Joanna Briggs Institute (JBI) and PRISMA-ScR guidelines.
• Extensive database searches across 13 sources with no language or time restrictions.
• Data synthesis using the PAGER framework by two independent reviewers.

Main Results:

• Nineteen studies utilized AI, primarily Machine Learning (ML) and Deep Learning, with Natural Language Processing (NLP) for data processing.
• ML models demonstrated superior predictive accuracy compared to traditional triage methods.
• AI applications focused on automated classification, severity prediction, and patient prioritization.

Conclusions:

• AI holds significant potential to enhance ED triage efficiency and clinical decision support.
• Current evidence is largely exploratory, with challenges in model performance consistency and external validation.
• Further research and robust validation are needed for safe, large-scale clinical implementation of AI in ED triage.