Assessment of the Diagnostic Performance and Clinical Impact of AI in Hepatic Steatosis: Systematic Review and Meta-Analysis

AI demonstrates remarkable potential for noninvasive screening and assessment of HS, especially in primary care. Nonetheless, clinical translation remains limited by substantial performance variability, predominance of retrospective designs, lack of rigorous external validation, and practical barriers such as data privacy and workflow integration. Future studies should prioritize prospective multicenter trials, standardized development pipelines, and robust external validation to bridge the gap between current evidence and clinical application. The key innovation of this review lies in establishing a unified, modality-agnostic analytical framework that integrates evidence beyond single-modality evaluations.