Artificial Intelligence in Trauma Care: A Systematic Review of Resuscitation, Diagnosis, Risk Prediction, and Management

BACKGROUND: Artificial intelligence (AI) is increasingly being explored in trauma care as a tool to support clinical decision-making. OBJECTIVE: To evaluate current evidence on AI applications in trauma resuscitation, diagnosis, complication prediction, and patient management. METHODS: A systematic review was conducted using PubMed, Web of Science, ScienceDirect, and Cochrane databases to identify studies published between 2015 and 2025. After screening 2968 records, 58 studies met inclusion criteria and were analyzed using a narrative synthesis approach. RESULTS: AI models were applied to trauma resuscitation (n = 8), diagnosis (n = 18), complication prediction (n = 23), and patient management (n = 9). Machine learning and deep learning approaches demonstrated performance comparable with, or exceeding, traditional clinical tools in some studies, including tasks such as predicting massive transfusion, detecting intracranial hemorrhage, forecasting mortality, and supporting intensive care unit or discharge planning. Interpretation of these findings is limited by the predominance of single-center, retrospective study designs and the frequent absence of external validation. CONCLUSIONS: AI shows meaningful potential to support clinical decision-making across multiple stages of trauma care, particularly in time-sensitive, high-risk settings. At present, the evidence base primarily reflects early-phase model development rather than clinical implementation. Future research should prioritize prospective, multicenter studies with external validation and evaluation of real-world clinical impact to determine whether AI-assisted decision support can improve patient-centered outcomes and safely integrate into trauma care workflows.