Machine learning techniques in hepatic encephalopathy: a scoping review

INTRODUCTION: Hepatic encephalopathy (HE) is defined as a specific type of cerebral dysfunction that encompasses a wide range of cognitive, psychomotor, and psychiatric disturbances. The burgeoning field of Artificial Intelligence (AI), particularly Machine Learning (ML), offers promising avenues for early detection and enhanced control of HE. This scoping review aims to provide a consolidated overview of AI’s role in the diagnosis and management of HE, thereby informing and guiding future research endeavors in this domain. METHODS: We followed Arksey and O’Malley’s methodological framework to perform this scoping review, using PubMed, Web of Science, Scopus, ScienceDirect, and IEEE databases to find relevant articles. We also utilized the PRISMA standard to report our review in a standardized manner. Studies that focused on the applications of AI or ML techniques in relation to the prediction or diagnosis of HE disease were included. RESULTS: Out of the 231 articles identified, 20 were ultimately included in this scoping review. The integration of artificial neural networks and expert systems represented an early and pioneering approach in applying AI to HE. Among supervised learning algorithms, Support Vector Machine emerged as the most frequently employed technique in HE research, based on our review of the selected studies. Notably, the primary application of AI in HE studies has been predictive modeling (n = 14), followed by five studies focused on classifying HE stages and one study analyzing patient survival using AI methodologies. CONCLUSIONS: This scoping review highlights the growing use of AI and ML diagnostic models and predictive tools utilizing various data types. These advancements have the potential to positively impact patient outcomes. Future research should focus on validating and implementing these AI models in clinical settings to assess their real-world effectiveness in improving patient care.