Enhancing Mass Casualty Triage Training Through Human-AI Collaboration
Disciplines
Human Factors Psychology
Abstract (300 words maximum)
In mass casualty events, first responders must quickly assess the health conditions of many victims so that those with the most severe injuries receive priority treatment. This process is called mass casualty triaging. Traditional triage training requires realistic mass casualty scenarios to ensure trainees can practice in environments that mimic real-world conditions. However, creating such scenarios for live training is resource-intensive in both time and cost. To address this, game-based simulation has emerged as an effective alternative, showing training outcomes comparable to traditional live exercises. To further enhance the effectiveness of game-based simulation, this study tested a new strategy in which participants learned to triage patients while collaborating with a robot partner. Participants watched a brief training video and then played a triage simulation game in which they assessed ten victims. Participants were randomly assigned to a control group or a collaboration group. In the control condition, the robot only followed participants in the game, while in the collaboration condition, the robot actively worked with them. Training effectiveness was measured through pre- and post-tests. We examined whether participants in the collaboration group achieved greater training effectiveness, showed higher reliance on the AI teammate, especially for difficult cases. We also looked at whether reliance on AI predicted improved performance. Findings are expected to provide insights into how human-AI collaboration can enhance learning, problem-solving, and preparedness in safety-critical scenarios.
Use of AI Disclaimer
no
Academic department under which the project should be listed
RCHSS – Psychological Science
Primary Investigator (PI) Name
Hansol Rheem
Enhancing Mass Casualty Triage Training Through Human-AI Collaboration
In mass casualty events, first responders must quickly assess the health conditions of many victims so that those with the most severe injuries receive priority treatment. This process is called mass casualty triaging. Traditional triage training requires realistic mass casualty scenarios to ensure trainees can practice in environments that mimic real-world conditions. However, creating such scenarios for live training is resource-intensive in both time and cost. To address this, game-based simulation has emerged as an effective alternative, showing training outcomes comparable to traditional live exercises. To further enhance the effectiveness of game-based simulation, this study tested a new strategy in which participants learned to triage patients while collaborating with a robot partner. Participants watched a brief training video and then played a triage simulation game in which they assessed ten victims. Participants were randomly assigned to a control group or a collaboration group. In the control condition, the robot only followed participants in the game, while in the collaboration condition, the robot actively worked with them. Training effectiveness was measured through pre- and post-tests. We examined whether participants in the collaboration group achieved greater training effectiveness, showed higher reliance on the AI teammate, especially for difficult cases. We also looked at whether reliance on AI predicted improved performance. Findings are expected to provide insights into how human-AI collaboration can enhance learning, problem-solving, and preparedness in safety-critical scenarios.