Table 1. Summary of Logistic Model Results by Using Different Parametersa.
| Prediction model (Th) | Sensitivity | Specificity | PPV | NPV | AUC (95% CI) | Rtrain (p value) | Deviance |
|---|---|---|---|---|---|---|---|
| SC (0.65) | 85.7% | 100% | 100% | 96.3% | 0.984 (0.934:1.0) | 0.911 (<0.0001) | 5.94 |
| PR (0.30) | 42.9% | 100% | 100% | 86.7% | 0.607 (0.302:0.885) | 0.212 (0.237) | 33.14 |
| PRR (0.38) | 85.7% | 92.3% | 75.0% | 96.0% | 0.907 (0.720:1.0) | 0.747 (<0.0001) | 18.92 |
| SC + PR (0.50) | 100% | 100% | 100% | 100% | 1.000 (1.0:1.0) | 1.000 (<0.0001) | 6.22e-13 |
| SC + PRR (0.50) | 100% | 100% | 100% | 100% | 1.000 (1.0:1.0) | 1.000 (<0.0001) | 1.94e-14 |
| PR + PRR (0.18) | 100% | 92.3% | 77.0% | 100% | 0.973 (0.912:1.0) | 0.774 (<0.0001) | 13.30 |
| SC + PR + PRR (0.50) | 100% | 100% | 100% | 100% | 1.000 (1.0:1.0) | 1.000 (<0.0001) | 1.60e-14 |
aSC: scattering coefficient, PR: phase retardation, PRR: phase retardation rate, Th: threshold.