Artificial Intelligence in Viral Hepatitis: Promise, Reality, And Barriers to Clinical Translation
Keywords:
Artificial intelligence, Machine learning, Viral hepatitis, Hepatitis B, Hepatitis C, Clinical translation, Clinical decision support systems, Fibrosis staging, Hepatocellular carcinoma prediction, External validation, Randomized controlled trials, Low- and middle-income countries, Health equity, Implementation science, WHO hepatitis eliminationAbstract
Background: Viral hepatitis affects an estimated 311 million people worldwide, with the greatest burden falling on low- and middle-income countries (LMICs). Artificial intelligence (AI) has emerged as a promising tool to improve screening, diagnosis, risk stratification, and clinical management. However, its real-world clinical utility and readiness for implementation remain uncertain. Methods: We conducted a comprehensive narrative review of artificial intelligence and machine-learning applications in chronic hepatitis B (HBV) and hepatitis C (HCV) across the continuum of care. Evidence was critically appraised with emphasis on methodological quality, external validation, clinical applicability, and equitable implementation, informed by TRIPOD+AI and SPIRIT-AI principles. Results: Across more than 50 studies, AI demonstrated strong technical performance (AUC approximately 0.80–0.95) for case-finding, fibrosis staging, and hepatocellular carcinoma risk prediction. The Intelligen-C clinical decision support system remains the only reported hepatitis AI tool with documented real-world implementation, demonstrating substantial improvements in HCV case-finding and healthcare efficiency. Nevertheless, external validation was uncommon, evidence from LMICs remained limited, prospective randomized trials were lacking, and economic evaluations were scarce. Important evidence gaps also persist in pediatric hepatitis, hepatitis D virus infection, and HBV/HIV or HCV/HIV coinfection.
