Table 3 Comparison of the diagnostic performances among models
cohort | Model 1 | Model 2 | Model 3 | |
|---|---|---|---|---|
Training set | AUC (95% CI) | 0.795(0.705–0.867) | 0.771(0.679–0.848) | 0.849(0.766–0.911) |
SPE (95% CI) | 0.604(0.460–0.733) | 0.698(0.555–0.813) | 0.849(0.719–0.928) | |
SEN (95% CI) | 0.865(0.736–0.940) | 0.750(0.608–0.855) | 0.673(0.528–0.793) | |
ACC (95% CI) | 0.733(0.730–0.737) | 0.724(0.720–0.728) | 0.762(0.759–0.765) | |
PPV (95% CI) | 0.682(0.554–0.788) | 0.709(0.569–0.820) | 0.814(0.661–0.911) | |
NPV (95% CI) | 0.821(0.659–0.919). | 0.740(0.594–0.849) | 0.726(0.596–0.828) | |
cut-off | 0.435 | 0.454 | 0.556 | |
Test set | AUC (95% CI) | 0.775(0.574–0.912) | 0.769(0.568–0.908) | 0.830(0.636–0.946) |
SPE (95% CI) | 0.692(0.389–0.896) | 0.846(0.537–0.973) | 0.846(0.537–0.973) | |
SEN (95% CI) | 0.929(0.642–0.996) | 0.643(0.356–0.860) | 0.786(0.488–0.943) | |
ACC (95% CI) | 0.815(0.804–0.826) | 0.741(0.727–0.755) | 0.815(0.804–0.826) | |
PPV (95% CI) | 0.765(0.498–0.922) | 0.818(0.478–0.968) | 0.846(0.537–0.973) | |
NPV (95% CI) | 0.900(0.541–0.995) | 0.688(0.415–0.879) | 0.786(0.488–0.943) | |
External validation set | AUC (95% CI) | 0.694(0.594–0.783) | 0.718(0.619–0.803) | 0.781(0.687–0.858) |
SPE (95% CI) | 0.600(0.452–0.733) | 0.800(0.659–0.895) | 0.700(0.552–0.817) | |
SEN (95% CI) | 0.740(0.594–0.850) | 0.600(0.452–0.733) | 0.780(0.637–0.880) | |
ACC (95% CI) | 0.670(0.666–0.674) | 0.700(0.696–0.704) | 0.740(0.736–0.744) | |
PPV (95% CI) | 0.649(0.511–0.768) | 0.750(0.585–0.868) | 0.722(0.581–0.831) | |
NPV (95% CI) | 0.698(0.537–0.823) | 0.667(0.532–0.780) | 0.761(0.609–0.869) |