Hotspots evolution and trend analysis of artificial intelligence applied in hepatocellular carcinoma since 2012: a bibliometric analysis

Purpose This study aims to analyze the evolution of hotspots and trends of artificial intelligence (AI) applied in hepatocellular carcinoma (HCC) using a visual bibliometric approach. Materials and methods We searched the Web of Science Core Collection database to select the studies related to AI applied in HCC from January 1, 2012, to March 15, 2023. The countries/regions, institutes, authors, journals, references, and keywords were visualized and analyzed. Results 961 studies were selected in this study. All studies were from 52 countries/regions, with China ( n = 585) leading the total number of publications, followed by the United States of America ( n = 167). These studies were published by 1518 institutions, mainly by Fudan University and Sun Yat-sen University with 56 and 55 studies, respectively. There are 370 journals related to the application of AI in HCC. In our study, we found that “expression”, “biomarker”, and “resection” had higher centrality of 0.48, 0.47, and 0.38, respectively. And 18 clusters were obtained, in which the Top 3 clusters in order are “#0 hepatocellular carcinoma”, “#1 deep learning”, and “#2 microvascular invasion”. After burst analysis, we found that “classification” was the strongest burst and the earliest emerging research direction. Conclusion AI was mainly applied in the field of HCC for tumor classification and microvascular invasion to construct prediction models using histopathological slices, medical images, and clinical features. It may be a key trend to improve the generalizability and interpretability of these models in the future.