Skip to main content

Table 2 Performance of proposed and existing prognostic tools for ICC

From: Development and validation of a gradient boosting machine to predict prognosis after liver resection for intrahepatic cholangiocarcinoma

Prognostic tools

C-statistic (95% CI)

P-value

Training/validation cohort (n = 401)

  GBM model

0.751 (0.717–0.784)

ref

  AJCC 8th edition

0.673 (0.637–0.708)

 < 0.001

  MEGNA prognostic score

0.674 (0.638–0.710)

 < 0.001

Test cohort (n = 649)

  GBM model

0.723 (0.697–0.749)

ref

  AJCC 8th edition

0.636 (0.608–0.664)

 < 0.001

  MEGNA prognostic scorea

0.617 (0.582–0.651)

 < 0.001

  1. Abbreviations: ICC intrahepatic cholangiocarcinoma, CI confidence intervals, GBM gradient boosting machine, AJCC American Joint Committee on Cancer, MEGNA multifocality, extrahepatic extension, grade, nodal status, and age, FAHNJMU First Affiliated Hospital of Nanjing Medical University, SEER Surveillance, Epidemiology, and End Results
  2. aAvailable at baseline (467/649) and compared with GBM model in corresponding sub-cohort