Table 3 Predictive performance for RFS of the proposed models
Models | Training dataset N = 129 | Validation dataset N = 55 |
|---|---|---|
C-index (95%CI) | C-index (95%CI) | |
Clinical model | ||
 clinicopathologic feature | 0.649 (0.592–0.706) | 0.556 (0.471–0.641) |
Radiomics model | ||
 Arterial phase | 0.767 (0.702–0.832) | 0.694 (0.623–0.832) |
Portal vein phase | 0.757 (0.692–0.821) | 0.736 (0.632–0.841) |
 Parenchymal phase | 0.789 (0.723–0.853) | 0.686 (0.582–0.791) |
All phases | 0.791 (0.726–0.856) | 0.690 (0.586–0.795) |
Combined model | ||
 Arterial phase + Clinicopathologic feature | 0.797 (0.732–0.862) | 0.732 (0.628–0.837) |
 Portal vein phase + Clinicopathologic feature | 0.792 (0.727–0.857) | 0.755 (0.651–0.860) |
 Parenchymal phase + Clinicopathologic feature | 0.806 (0.741–0.871) | 0.728 (0.624–0.834) |
 All phases + Clinicopathologic feature | 0.809 (0.744–0.874) | 0.724 (0.620–0.829) |