Table 4 Comparison of the CR model and CRR model in three set
From: Added value of CE-CT radiomics to predict high Ki-67 expression in hepatocellular carcinoma
Models | Training set | Internal test set | External validation set | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Models | AUC value (95%CI) | Sensitivity (%) | Specificity (%) | Accuracy (%) | AUC value (95%CI) | Sensitivity (%) | Specificity (%) | Accuracy (%) | AUC value (95%CI) | Sensitivity (%) | Specificity (%) | Accuracy (%) |
CR | 0.836(0.765-0.907) | 86 | 70 | 78.33 | 0.805(0.683-0.926) | 90 | 71 | 70.59 | 0.805(0.657-0.953) | 76 | 75 | 75.68 |
CRR | 0.903(0.849-0.956) | 86 | 79 | 82.5 | 0.848(0.742-0.954) | 83 | 76 | 58.82 | 0.877(0.676-0.986) | 64 | 70 | 62.16 |
Rad-score | 0.762(0.673-0.850) | 87 | 67 | 73.33 | 0.687(0.536-0.839) | 74 | 65 | 47.06 | 0.667(0.495-0.839) | 48 | 66 | 37.84 |