Clinical Application of Artificial Intelligence in the Management of Chronic Hepatitis B

Chronic hepatitis B (CHB) burdens the healthcare system, affecting millions of individuals worldwide with a potential progression to complications such as cirrhosis and hepatocellular carcinoma (HCC). The World Health Organization launched the chronic viral hepatitis elimination plan which aimed to eliminate chronic hepatitis B virus by 2030, yet the goal has seemed to be far from reach with the remaining few years owing to different hurdles ranging from screening strategy and diagnosis to risk stratification for antiviral treatment and HCC surveillance. To refine the management of CHB, previous studies focused on identifying the at-risk populations and predictors for adverse outcomes using conventional statistical methods. With the emerging role of artificial intelligence (AI), there is further room to improve different aspects of screening and management. For instance, generative AI, such as large language models (LLMs), has been proven to have the ability to answer different clinical questions, assisting patients' and physicians' decision on various clinical scenarios. Different machine learning and deep learning models have also been developed to assess for liver fibrosis, as well as to aid segmentation of liver structures and detection of focal liver lesions through various imaging modalities. While different AI models have shown superior performance compared to those developed through traditional statistical models, the lack of standardized datasets in training cohorts and the unresolved concerns of hallucination of false information in LLMs pose challenges on applying AI in the current management of CHB, in which further studies are required to adopt AI into daily clinical practice.