Table 4.
Effect on sensitivity, specificity, PPV and NPV at the respective cutoff of different models.
| Variables of models | Models with different features | |||
|---|---|---|---|---|
| Model A | Model B | Model C | Model D | |
| AUC of training set | 0.953(0.911~0.996) | 0.903(0.844~0.962) | 0.947(0.901~0.992) | 0.957(0.916~0.999) |
| Cutoff | 0.408 | 0.497 | 0.498 | 0.412 |
| Sensitivity(%) | 92.0(81.2~96.9) | 76.0(62.6~85.7) | 92.0(81.2~96.9) | 94.0(83.8~97.9) |
| Specificity(%) | 94.8(85.9~98.2) | 96.6(88.3~99.1) | 93.1(83.6~97.3) | 93.1(83.6~97.3) |
| PPV (%) | 93.9(83.5~97.9) | 95.0(84.5~98.6) | 92.0(81.2~96.9) | 92.2(81.5~96.9) |
| NPV (%) | 93.2(83.8~97.3) | 82.4(71.6~89.6) | 93.1(83.6~97.3) | 94.7(85.6~98.2) |
Features in different models as follows:
Model A: age+ Onecut2 methylation.
Model B: age+ gene mutations.
Model C: age+ panel test.
Model D: age+ TERT C228T/C250T+ Onecut2 methylation.
PPV, Positive predictive value; NPV, Negative predictive value.
95% CI values were showed in brackets.