AI-Assisted Triage in Disaster Medicine: Enhancing Response Time and Clinical Decision-Making Without Replacing Human Judgment

Abstract

The increasing frequency and severity of natural disasters have intensified the demands placed on emergency response systems, particularly in mass-casualty incidents where rapid and accurate triage is critical. Traditional disaster triage frameworks, including Simple Triage and Rapid Treatment (START) and Sort, Assess, Lifesaving Interventions, Treatment/Transport (SALT), provide structured methods for prioritizing patients under resource-limited conditions. However, their effectiveness may be compromised by environmental chaos, cognitive overload, limited situational awareness, and variability in clinical judgment. This study examines the extent to which artificial intelligence (AI)–assisted decision-support systems can enhance triage speed and accuracy while preserving clinician oversight and ethical accountability. An integrative literature review was conducted, synthesizing peer-reviewed research on disaster triage accuracy, under- and over-triage rates, cognitive factors affecting emergency decision-making, and emerging applications of AI in emergency medicine. A conceptual systems analysis compared traditional triage frameworks with AI-assisted models across key dimensions, including assessment speed, prioritization accuracy, cognitive burden, adaptability to hazardous environments, and ethical safeguards. To illustrate applied integration, a conceptual case study—MedResQ, a semi-autonomous, multi-terrain robotic platform equipped with thermal imaging, ultrasonic sensing, and machine learning–based human recognition—was evaluated as an example of AI-supported situational awareness within established triage workflows. Findings from the reviewed literature suggest that AI-assisted systems demonstrate potential to reduce under-triage, improve early identification of high-risk patients, and enhance situational awareness in environments where responder access is delayed or unsafe. Decision-support models in emergency departments show improved predictive performance compared to traditional scoring systems, while drone- and robotics-assisted assessment may accelerate victim localization during mass-casualty events. However, current evidence is largely derived from simulations, retrospective analyses, and pilot studies, highlighting the need for prospective field validation and standardized evaluation metrics. AI-assisted triage systems, when implemented within a human-in-the-loop framework that preserves clinician authority, may represent a meaningful augmentation of disaster medicine. Ethical safeguards, transparency, bias mitigation, and clear accountability structures remain essential for responsible deployment. Future research should focus on real-world testing, human–AI interaction under stress, and the development of interoperable protocols to ensure scalable and equitable implementation across diverse disaster contexts.

Keywords

Applied Science - Engineering

Citation

References

Reference metadata is pending and must be finalized before DOI deposit.