Additional file 1 of Artificial neural network model to predict post-hepatectomy early recurrence of hepatocellular carcinoma without macroscopic vascular invasion

Additional file 1: Supplementary Table 1.Staging systems of whole cohort, derivation cohort and validation cohort. Supplementary Table 2. The performance of the ANN model, Cox model and these eight prognostic factors individually in predicting post-hepatectomy early recurrence in derivation cohort. Supplementary Figure 1. Flow chart of the study design. Supplementary Figure 2. Recurrence-risk stratification curves of the eight prognostic factors in the derivation cohort. (A) HBV-DNA, (B) GGT) level, (C) AFP level, (D) tumor size), (E) tumor differentiation, (F) MVI, (G) satellite nodules, (H) blood loss. Abbreviations: HBV-DNA, Hepatitis B virus deoxyribonucleic acid load; GGT, γ-glutamyl transpeptidase; AFP, α-fetoprotein; MVI; microvascular invasion. Supplementary Figure 3. Screenshot of the SPSS software-based ANN model to predict PHER risk. First, SPSS is opened, and the values of the eight prognostic factors (HBV-DNA load, GGT level, AFP level, tumor size, tumor differentiation, MVI, satellite nodules, and blood loss) are entered into the program. Next, the Scoring Wizard on the toolbar is found, and the ANN model file is selected. The program will automatically calculate PHER risk. For ex ample, in an HCC patient with high levels of HBV-DNA, GGT, and AFP; large tumor size; poor tumor differentiation; high blood loss; and presence of MVI and satellite nodules, the program calculated a predicted PHER risk of 0.84. This patient was classified in the high-risk group based on the risk stratification. Abbreviations: HCC, hepatocellular carcinoma; ANN, artificial neural network; PHER, post-hepatectomy early recurrence; HBV-DNA, hepatitis B virus DNA load; GGT, γ-glutamyl transpeptadase; AFP, α-fetoprotein; MVI, microvascular invasion.