Artificial intelligence in advanced liver disease: From risk stratification to decision
TL;DR
Recent advances in AI technology regarding various clinical aspects of liver cirrhosis and portal hypertension are discussed, including the diagnosis of cirrhosis and clinically significant portal hypertension, detection and prediction of major cirrhosis-related complications, and assessing the prognosis.
Abstract
The diagnosis and management of liver cirrhosis and portal hypertension have been challenging for hundreds of years, placing a tremendous burden on medical and social resources. In routine practice, clinicians usually depend on various documented scoring and stratification systems to establish the diagnosis and make decisions. These systems have traditionally been constructed through manual review of retrospective data, image interpretation, and statistical analyses. However, the development of artificial intelligence (AI) has allowed for the analysis of large and varied datasets, and consequently the ability to identify novel factors and test the predictive power of different combinations of factors that would not be feasible using traditional means. Considering the promise of using AI for research in this notorious disease entity, this review article discusses recent advances in AI technology regarding various clinical aspects of liver cirrhosis and portal hypertension. Topics covered include the diagnosis of cirrhosis and clinically significant portal hypertension, detection and prediction of major cirrhosis-related complications (gastroesophageal varices, hepatic encephalopathy, infection, acute kidney injury, hepatorenal syndrome, and hepatocellular carcinoma), assessing the prognosis, and evaluating the treatment response. The potential benefits, drawbacks, limitations, and future perspectives of AI are also discussed.