Prehospital emergency care is a critical component of trauma management, as timely and accurate injury assessment significantly influences patient outcomes. Current trauma triage methods primarily rely on static scoring methods; however, real-world prehospital scenarios are often complex and dynamic, rendering existing approaches insufficient for adapting to rapidly changing conditions. To address this challenge, this paper proposes a novel trauma triage method capable of effectively capturing real-time variations in patients’ vital signs while emphasizing key frequency-domain features for accurate injury classification. Specifically, the method employs a Bidirectional Gated Recurrent Unit (BiGRU) to extract temporal features that reflect dynamic changes in vital sign sequences. In addition, a Frequency-domain Channel Attention Mechanism (FECAM) is introduced to identify critical features across different time intervals. Experimental results demonstrate the effectiveness and practical utility of the proposed approach.