Introduction: Triage and casualty evacuation are two essential components of the medical management of mass casualty incidents. However, there is a notable lack of scientific evidence supporting the validity of mass casualty triage protocols. This study aims to develop a ‘Drone Integrated, Artificial Intelligence-Based Mass Casualty Triage Decision Support System’ (DIABMaCTDeS) to assist on-scene response teams in making triage decisions and to enhance data transfer from the incident site to emergency management centers. Methods: The development of the system occurred in four phases: (i) respiratory rate, pulse, oxygen saturation and body temperature detection algorithms were created by utilizing image processing techniques applied to drone captured footage; (ii) Drone Integrated Mass Casualty Triage Algorithm (DIMaCTA) was developed using the Delphi method; (iii) an artificial intelligence-based decision support system was built upon the DIMaCTA algorithm; (iv) a user application was developed to facilitate real-time data transfer between field teams and emergency management centers. Results: According to the preliminary results, the accuracy rate of the pulse, respiration, oxygen saturation, and body temperature detection algorithms varied between 95% and 99%. In the Delphi study, the interquartile range for expert opinions on decision points and scoring criteria was determined to be 1.2, and the total percentage of those who answered “partially agree” and “strongly agree” regarding the structure of the algorithm was 70% or above. These results show that a consensus was reached on the parameters and decision points in the algorithm. Conclusion: The drone-integrated system enables faster scene access than ground teams, providing data on scene security, victim location, vital signs, and triage priorities. It allows the most urgent cases to be identified and evacuated quickly and safely. The system will significantly aid in the rapid assessment of casualties and data transfer to emergency management centers, both in urban and hard-to-access remote areas, especially where responder safety is uncertain.
Tayfur et al. (Sun,) studied this question.