Skip to main content

Table 3 The performance of the ANN model and other models and staging systems in predicting early recurrence in derivation cohort and training cohort

From: Artificial neural network model to predict post-hepatectomy early recurrence of hepatocellular carcinoma without macroscopic vascular invasion

Staging systems / models

Training cohort

Validation cohort

AUC

95% CI

P vaule

AUC

95% CI

P vaule

BCLC stage

0.536

0.495–0.648

< 0.05

0.572

0.495–0.648

< 0.05

TNM8th stage

0.503

0.458–0.549

< 0.05

0.481

0.405–0.558

< 0.05

Okuda stage

0.535

0.428–0.584

< 0.05

0.506

0.428–0.584

< 0.05

CNLC stage

0.599

0.593–0.741

< 0.05

0.667

0.593–0.741

> 0.05

HKLC stage

0.589

0.496–0.648

< 0.05

0.572

0.496–0.648

< 0.05

French stage

0.584

0.470–0.625

< 0.05

0.547

0.470–0.625

< 0.05

CLIP score

0.565

0.421–0.578

< 0.05

0.499

0.421–0.578

< 0.05

JIS score

0.504

0.376–0.529

< 0.05

0.453

0.376–0.529

< 0.05

Ng et al.’s model

0.609

0.573–0.721

< 0.05

0.647

0.573–0.721

< 0.05

Shim et al.’s model

0.619

0.517–0.667

< 0.05

0.592

0.517–0.667

< 0.05

CPH model

0.733

0.657–0.792

< 0.05

0.724

0.657–0.792

> 0.05

ANN model

0.753

0.668–0.803

Ref

0.736

0.668–0.803

Ref

  1. Abbreviations: BCLC Barcelona Clinic Liver Cancer, TNM 8th 8th edition of TNM /AJCC, CNLC China Liver Cancer, HKLC Hong Kong Liver Cancer, CLIP Cancer of the Liver Italian Program, JIS Japan Integrated Staging, CPH Cox’s proportional hazards, ANN artificial neural network