430 A Review of Artificial Intelligence Solutions for Hepatocellular Carcinoma

Abstract

Abstract Introduction Hepatocellular carcinoma (HCC) is a significant cause of morbidity and mortality worldwide. Despite significant advancements, the diagnosis and management of HCC remain a challenge. This review aims at exploring artificial intelligence (AI) solutions applied to HCC. Method A review of the literature from Embase, MEDLINE and Cochrane Library was conducted to determine the role of AI in HCC, across three domains: detection, characterisation, and prediction. 56 relevant original research studies were identified and included in a qualitative synthesis. Results AI models can be implemented into detection of HCC, as they excel at analysis and integration of large datasets. Moreover, AI outclasses traditional statistical models at tumour characterisation based on radiological and pathological images. Predicting treatment outcomes and survival using AI can shape future HCC guidelines and support clinical decision making, especially treatment choice. AI in HCC has limitations, hindering its clinical adoption. Small sample size, single-centre data, non-transparent reporting, lack of external validation, and overfitting all results in low generalisability of findings. Conclusions AI has immense potential; however, interdisciplinary collaboration is needed to improve, validate, and implement it across all aspects of HCC. AI has a multifaceted role in HCC and its importance can increase in the future, as more sophisticated technologies emerge.