Table 8. Margin classification results.
| Class code | Method | AUC (95% CI) | ACC (95% CI) | SN (95% CI) | SP (95% CI) | Phl | P value (AUC, ACC, SN, SP) |
|---|---|---|---|---|---|---|---|
| Mar1 | M1 | 0.7433 (0.71, 0.78) | 0.6990 (0.65, 0.75) | 0.6198 (0.54, 0.70) | 0.7642 (0.69, 0.84) | 0.26 | 0.04, 0.49, 0.65, 0.9 |
| M2 | 0.7522 (0.72, 0.79) | 0.7048 (0.65, 0.76) | 0.6296 (0.54, 0.72) | 0.7665 (0.69, 0.84) | 0.51 | ||
| Mar2 | M1 | 0.9335 (0.92, 0.94) | 0.8607 (0.83, 0.89) | 0.8691 (0.83, 0.91) | 0.8524 (0.81, 0.89) | 0.12 | 0.1, 0.01, 0.15, 0.05 |
| M2 | 0.9352 (0.92, 0.95) | 0.8699 (0.84, 0.90) | 0.8745 (0.83, 0.92) | 0.8653 (0.82, 0.91) | 0.61 | ||
| Mar3 | M1 | 0.8300 (0.81, 0.85) | 0.7560 (0.72, 0.79) | 0.7600 (0.69, 0.83) | 0.7515 (0.69, 0.81) | 0.31 | 0.0009, 0.001, 0.16, 0.0013 |
| M2 | 0.8371 (0.82, 0.86) | 0.7732 (0.73, 0.81) | 0.7696 (0.71, 0.83) | 0.7764 (0.71, 0.84) | 0.39 |
M1: equal-weighted ensemble; M2: diversity-weighted ensemble. AUC, area under the receiver operating characteristic curve; CI, confidence interval; Phl, Hosmer-Lemeshow test statistic; ACC, accuracy; SN, sensitivity; SP, specificity.