Integrated Multimodal AI for Emergency Triage Optimization

In critical care settings, timely and accurate triage is essential to prevent patient deterioration. This study presents an integrated multimodal AI framework that combines chest X-ray imaging with vital signs data to improve the accuracy and speed of emergency triage decisions. By processing visual and physiological inputs in parallel, the proposed model predicts both the patient's current condition and the estimated time to a potential failure event. Experimental evaluation demonstrates that the multimodal model significantly outperforms unimodal baselines, achieving over 90% classification accuracy and a low mean absolute error in time-to-failure predictions. These findings suggest that combining diverse data sources can substantially enhance the effectiveness of clinical decision support systems [1], [2].