Table 2.
Performance of Logistic Regression Model for NMIBC discrimination.
| AUC | Sensitivity | Specificity | |
|---|---|---|---|
| aNMIBC vs. CONTROL | |||
| Training/discovery | 0.857 (0.837~0.878) | 0.754 (0.722~0.787) | 0.786 (0.749~0.823) |
| 10-fold cross-validation | 0.833 (0.764~0.901) | 0.737 (0.737~0.836) | 0.792 (0.683~0.902) |
| External validation | 0.838 (0.769~0.953) | 0.807 (0.730~0.853) | 0.818 (0.769~0.940) |
| bLOW-GRADE NMIBC vs. CONTROL | |||
| Training/discovery | 0.938 (0.923~0.953) | 0.899 (0.869~0.930) | 0.791 (0.750~0.832) |
| 10-fold cross-validation | 0.899 (0.836~0.963) | 0.881 (0.881~0.979) | 0.786 (0.662~0.910) |
| cHIGH- AND LOW-GRADE NMIBC WITHOUT HEMATURIA | |||
| Training/discovery | 0.802 (0.770~0.834) | 0.751 (0.708~0.795) | 0.757 (0.711~0.803) |
| 10-fold cross-validation | 0.755 (0.645~0.866) | 0.762 (0.762~0.891) | 0.757 (0.619~0.895) |
| dHIGH- and LOW-GRADE NMIBC WITH HEMATURIA | |||
| Training/discovery | 0.878 (0.852~0.904) | 0.858 (0.804~0.912) | 0.668 (0.631~0.705) |
| 10-fold cross-validation | 0.827 (0.731~0.923) | 0.889 (0.889~1.000) | 0.667 (0.555~0.778) |
a
The panel: Dopamine 4-sulfate, MG00/1846Z,9Z,12Z,15Z/00, Aspartyl-Histidine, Tyrosyl-Methionine.
b
The panel: 3-Hydroxy-cis-5-tetradecenoylcarnitine, 6-Ketoestriol, Beta-Cortolone, Tetrahydrocorticosterone, Heptylmalonic acid.
c
The panel: N-Acetyl-4-O-acetylneuraminic acid, 4-(2-Aminophenyl)-2,4-dioxobutanoic acid, 6-Keto-decanoylcarnitine, 3-hydroxydecanoyl carnitine, 2-Hydroxylauroylcarnitine.
d
The panel: Indolylacryloylglycine, Histidinyl-Histidine, Indoleacrylic acid, N-acetyl-5-methoxykynuramine, L-3-Hydroxykynurenine.