LLM use in escape room trauma simulations

This randomized controlled study investigates the impact of large language model (LLM) assistance—specifically ChatGPT—on decision-making, teamwork, and confidence in escape room based trauma simulations. Forty undergraduate medical students at Vaasa Central Hospital, University of Turku, were randomized into LLM-assisted or non-LLM teams. Participants completed 18 structured trauma scenarios involving airway obstruction, pneumothorax, orthopedic trauma, and hemorrhage control. Observed outcomes included decision-making speed and accuracy, teamwork quality, and changes in confidence. The study also assessed knowledge retention and participant experiences with AI.