Table 3.
Discriminative information comparison using logistic regression models with MS status as the outcome.
|
Metrics of model
discriminability |
Model comparison | ||||
|---|---|---|---|---|---|
|
Models
(Predictors) |
AUROC (95% CI) |
Likelihood
ratio χ2 |
Comparison pair |
p
-value by Delong's
test |
p
-value by likelihood
ratio test |
| A: (6MWGST + 6MWTD + age + sex) |
0.93 (0.86–0.99) | 38.3 | Model A against Model C | 0.037 | 0.002 |
| B: (Δ6MW + 6MWTD + age + sex) |
0.85 (0.76–0.94) | 25.8 | Model B against Model C | 0.491 | 0.109 |
| C: (6MWTD + age +sex) |
0.83 (0.73–0.93) | 23.2 | Model C against Model E | 0.058 | <0.001 |
| D: (T25FW + age + sex) |
0.78 (0.65–0.90) | 15.6 | Model D against Model E | 0.129 | 0.006 |
| E: (age + sex) |
0.70 (0.54–0.86) | 8.2 | Model A against Model B | 0.045 | N/A |