Abstract
Intracranial hypertension (ICH) is a critical complication of traumatic brain injury (TBI), associated with poor outcomes. AI shows promise for early ICH prediction, but its clinical integration remains uncertain. This systematic review evaluates the performance, clinical applicability, and limitations of AI models for ICH prediction in TBI. We searched PubMed, Embase, IEEE Xplore, and Scopus, identifying 250 records. After removing duplicates and screening titles and abstracts, 37 full-text articles were assessed, with 9 studies meeting the inclusion criteria. Risk of bias was evaluated using PROBAST, and data on algorithms, performance metrics, and clinical integration were extracted. The included studies demonstrated strong predictive performance, with ensemble models achieving the highest accuracy. However, reliance on invasive monitoring, small sample sizes, and retrospective designs limited generalizability. Only one non-AI study reported clinical integration, highlighting a translational gap. While AI models show potential for ICH prediction, methodological heterogeneity and the lack of prospective validation hinder clinical adoption. Future research should prioritize standardized outcomes, model explainability, and real-world testing to bridge this gap.