Table 3.
External Validation (Dataset 1) Receiver Operating Characteristic Curves and AUC Analysis for the SP A1 Parameter, CBI, and TBI Using Naive Bayes Classifier and for the Processed Pseudoimages Using DenseNet 121
| Parameter (Pairwise Comparison) | Mean Δ AUC ± SE | 95% CI | Z Statistic | P Value |
|---|---|---|---|---|
| SP A1 and DenseNet121 | 0.020 ± 0.024 | 0.0432, 0.0832 | 0.83 | 0.198 |
| CBI and DenseNet121 | 0.048 ± 0.021 | 0.0020, 0.0980 | 2.28 | 0.013 |
| TBI and DenseNet121 | 0.066 ± 0.026 | 0.0211, 0.1108 | 2.53 | 0.006 |
| SP A1 and CBI | 0.028 ± 0.013 | 0.0130, 0.0690 | 2.15 | 0.016 |
| SP A1 and TBI | 0.046 ± 0.020 | 0.0115, 0.0804 | 2.30 | 0.009 |
| CBI and TBI | 0.018 ± 0.010 | 0.0007, 0.0352 | 2.00 | 0.021 |
Δ AUC = difference between the area under the curve; CBI = Corvis biomechanical index; CI = confidence interval; ROC = receiver operating characteristic; SE = standard error; SP A1 = stiffness parameter at first applanation; TBI = tomography and biomechanical index.
The bold type signifies P ≤ 0.05.