TABLE 5.
Statistics of regression models for survival predictions
| Model (mo) by type | Variable (mo)a | Intercept (P value)b | Parameter (P value)b | AUC (95% CI)c | P value for AUC comparison |
|---|---|---|---|---|---|
| Logistic | |||||
| Peak anti-PA IgG | IgG_7 | −0.6569 (0.0582) | 1.0009 (<0.0001) | 0.7956 (0.7208–0.8703) | 0.2712 |
| Last anti-PA IgG | IgG_last | 0.7105 (0.0015) | 2.1628 (<0.0001) | 0.8214 (0.7514–0.8914) | Reference model |
| Last anti-PA IgG (12) | IgG_last (12) | 0.5416 (0.0707) | 1.4605 (0.0049) | 0.7240 (0.5970–0.8509) | NAd |
| Last anti-PA IgG (30) | IgG_last (30) | 1.0465 (0.0269) | 2.3380 (0.0036) | 0.8758 (0.7721–0.9796) | NA |
| Last anti-PA IgG (52) | IgG_last (52) | 0.3651 (0.5105) | 4.4283 (0.0026) | 0.9071 (0.8014–1.000) | NA |
| Peak TNA | TNA_7 | −1.9889 (0.0006) | 1.0808 (<0.0001) | 0.7918 (0.7158–0.8678) | 0.2651 |
| Last TNA | TNA_last | −1.1918 (0.0086) | 1.6011 (<0.0001) | 0.7270 (0.6526–0.8014) | 0.0009e |
| SI_2 | SI_2 | −4.9416 (<0.0001) | 1.4900 (<0.0001) | 0.7860 (0.7086–0.8633) | 0.3873 |
| SI_2 + peak anti-PA IgG | SI_2 | −2.8355 (0.0553) | 0.6652 (0.1390) | 0.8121 (0.7412–0.8831) | 0.7435 |
| IgG_7 | 0.7267 (0.0136) | ||||
| SI_2 + last anti-PA IgG | SI_2 | −1.6413 (0.2724) | 0.6155 (0.1140) | 0.8424 (0.7758–0.9090) | 0.0777 |
| IgG_last | 1.7901 (0.0001) | ||||
| SI_2 + peak TNA | SI_2 | −3.9897 (0.0024) | 0.7455 (0.0918) | 0.8100 (0.7384–0.8817) | 0.7115 |
| TNA_7 | 0.7368 (0.0197) | ||||
| SI_2 + last TNA | SI_2 | −4.3354 (0.0006) | 1.0212 (0.0057) | 0.7927 (0.7182–0.8672) | 0.3462 |
| TNA_last | 0.9827 (0.0224) | ||||
| Peak anti-PA IgG + last anti-PA IgG | IgG_7 | 0.4400 (0.3400) | 0.1991 (0.5027) | 0.8247 (0.7546–0.8947) | 0.5252 |
| IgG_last | 1.8750 (0.0012) | ||||
| Peak TNA + last anti-PA IgG | TNA_7 | 0.2292 (0.7834) | 0.1928 (0.5497) | 0.8266 (0.7570–0.8963) | 0.3859 |
| IgG_last | 1.9140 (0.0008) | ||||
| Peak anti-PA IgG + last TNA | IgG_7 | −1.1664 (0.0116) | 0.7297 (0.0037) | 0.7942 (0.7202–0.8682) | 0.2058 |
| TNA_last | 0.7490 (0.1054) | ||||
| Peak TNA + last TNA | TNA_7 | −2.1130 (0.0003) | 0.7725 (0.0058) | 0.7956 (0.7217–0.8695) | 0.2639 |
| TNA_last | 0.7589 (0.1059) | ||||
| PCLR | |||||
| SI_2 + peak anti-PA IgG + peak TNA + last anti-PA IgG | PC1 | 1.1927 (<0.0001) | −0.0356 (<0.0001) | 0.8403 (0.7744–0.9062) | 0.1780 |
| PC2 | 0.5438 (0.0076) | ||||
| PC3 | 1.7489 (0.0046) | ||||
| PC4 | −0.0977 (09300) | ||||
| SI_2 + peak anti-PA IgG + last TNA + peak TNA + last anti-PA IgG | PC1 | 1.1510 (<0.0001) | −0.0346 (<0.0001) | 0.8432 (0.7768–0.9097) | 0.6185 |
| PC2 | 0.5254 (0.0091) | ||||
| PC3 | 0.7204 (0.1153) | ||||
| PC4 | 2.1960 (0.0148) | ||||
| PC5 | 0.9948 (0.4113) |
a
PC1, PC2, PC3, PC4, and PC5 are principal components.
b
Intercept and parameter values are for the logistic regression for the indicated variable or component. Note that models with multiple variables have only a single intercept.
c
AUC is given as a measure of model accuracy.
d
NA, not applicable, due to dependent ROC curve comparisons being performed in this analysis, and thus ROC curve comparisons here should be within exactly the same sample.
e
P value with a significance level of <0.05 for AUC comparison with that of the logistic regression model with last anti-PA IgG only.