A machine learning approach to identify optimal candidates for transarterial chemoembolization in unresectable HBV-related hepatocellular carcinoma complicated by first-branch portal vein tumor thrombus: a multicenter study

The integrated application of RSF and PSM can effectively identify low-risk candidates for TACE among patients with unresectable HCC and PVTT1. Our findings provide strong evidence that for this carefully selected patient subgroup, TACE offers survival outcomes comparable to those for patients without PVTT, highlighting the clinical utility of machine learning in guiding treatment decisions for this challenging disease.