Table 2.
Predictive performance of histological grading models.
| Models | Sensitivity (95% CI) | Specificity (95% CI) | Accuracy (95% CI) | AUC (95% CI) |
|---|---|---|---|---|
| Radiomics score | ||||
| training | 52 of 79 (0.54, 0.76) | 91 of 121 (0.66, 0.83) | 143 of 200 (0.65, 0.78) | 0.78 (0.71,0.84) |
| test | 27 of 33 (0.65, 0.93) | 33 of 51 (0.50, 0.78) | 60 of 84 (0.61, 0.81) | 0.75 (0.64,0.85) |
| validation | 10 of 10 (0.69, 1.00) | 16 of 32 (0.32, 0.68) | 26 of 42 (0.32, 0.65) | 0.76 (0.60,0.91) |
| Clinical model | ||||
| training | 56 of 79 (0.60, 0.81) | 69 of 121 (0.48, 0.66) | 125 of 200 (0.55, 0.69) | 0.67 (0.60,0.75) |
| test | 24 of 33 (0.55, 0.87) | 29 of 51 (0.42, 0.70) | 53 of 84 (0.52, 0.73) | 0.68 (0.56,0.79) |
| Nomogram | ||||
| training | 68 of 79 (0.76, 0.93) | 71 of 121 (0.49, 0.68) | 139 of 200 (0.63, 0.76) | 0.80 (0.74,0.86) |
| test | 29 of 33 (0.72, 0.97) | 32 of 51 (0.48,0.76) | 61 of 84 (0.63, 0.95) | 0.77 (0.66,0.87) |
AUC, area under the receiver operating characteristic (ROC) curve.