Skip to main content

Table 3 Diagnostic Performances of All Methods for Predicting Liver Cirrhosis in the training and validation cohort

From: A radiomics-based model on non-contrast CT for predicting cirrhosis: make the most of image data

  Training (n = 144) Validation (n = 150) Training vs. Validation
Methods AUROC (95%CI) AUROC (95%CI) Delong test
Radiomics nomogram 0.915 (0.869, 0.961) 0.872 (0.814, 0.930) P = .257
CT-reported cirrhosis status 0.752 (0.683, 0.821) 0.755 (0.683, 0.827) P = .961
APRI 0.725 (0.642, 0.809) 0.731 (0.649, 0.814) P = .921
FIB-4 0.664 (0.575, 0.753) 0.688 (0.601, 0.775) P = .705
Comparison of AUROC (Delong test)
Radiomics nomogram vs. CT-reported cirrhosis status P < .001 P = .006  
Radiomics nomogram vs. APRI P < .001 P = .003  
Radiomics nomogram vs. FIB-4 P < .001 P < .001  
CT-reported cirrhosis status vs. APRI P = .594 P = .651  
CT-reported cirrhosis status vs. FIB-4 P = .073 P = .201  
APRI vs. FIB-4 P = .040 P = .040  
  1. Note. ——Data in parentheses are the 95% confidence interval. APRI aspartate transaminase-to-platelet ratio index, AUROC area under the receiver operating characteristic, FIB-4 fibrosis-4