Table 4.
Accuracy of the prediction score of the nomogram for evaluating the risk of PHLF grade B/C incidence.
| Variable | Training set (n = 383) | Validation set (n = 191) | ||
|---|---|---|---|---|
| Value | 95% CI | Value | 95% CI | |
| Area under ROC curve | 0.832 | 0.777–0886 | 0.803 | 0.723–0.883 |
| Cut-off score | 2.109 | — | 1.66 | — |
| C-index | 0.832 | — | 0.808 | — |
| Sensitivity (%) | 0.813 | 0.690–0.903 | 0.884 | 0.698–0.975 |
| Specificity (%) | 0.731 | 0.679–0.778 | 0.618 | 0.539–0.692 |
| Positive predictive value (%) | 35.5 | 30.0–54.0 | 26.7 | 20.8–65.5 |
| Negative predictive value (%) | 95.5 | 91.6–96.5 | 97.1 | 91.1–97.9 |
| Positive likelihood ratio | 3.02 | 2.43–3.76 | 2.31 | 1.82–2.94 |
| Negative likelihood ratio | 0.25 | 0.14–0.43 | 0.18 | 0.06–0.54 |