Abstract
OBJECTIVE: Posthepatectomy recurrence of hepatocellular carcinoma (HCC) is a major cause of poor prognosis. Accurate prediction is essential for reducing the burden of advanced disease and improving outcomes. METHODS: A systematic search of the PubMed, Embase, and Cochrane Library databases was conducted from their inception to December 31, 2024. The standard quality assessment of diagnostic accuracy studies (QUADAS-2) tool was utilized to analyse the methodological quality of the included studies. Bivariate linear mixed models were used to pool diagnostic estimates, including sensitivity (Se), specificity (Sp), positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR). Additionally, the area under the receiver operating characteristic curves (AUC) of the included studies was utilized to evaluate the diagnostic value. RESULTS: A total of 6665 HCC patients included in 20 studies were enrolled. The pooled Se, Sp, PLR, NLR, DOR and AUC for the overall AI-assisted diagnostic performance for postoperative HCC recurrence were 0.87 (95% CI: 0.72-0.83), 0.85 (95% CI: 0.80-0.90), 5.39 (95% CI: 3.85-7.55), 0.25 (95% CI: 0.20-0.33), 21 (95% CI: 13-35), and 0.89 (95% CI: 0.86-0.91), respectively. CONCLUSION: AI showed high accuracy in predicting the posthepatectomy recurrence of HCC and would shed light on screening and monitoring high-risk patients following liver resection for further treatment.