Table 3.
Logistic regression models and their associated sensitivity, specificity and AUC
| A. Predictive Models (Visit 1) | Predictors | OR | P | 95% Conf. Int | |
|---|---|---|---|---|---|
| i. First 15 s (GP-MCHAT-R model) | Social-15 s | 0.85 | 0.001 | 0.78 | 0.94 |
| Model LR = − 37.96, AIC = 0.811 | MCHAT-R | 21.6 | < 0.001 | 4.5 | 103.83 |
| constant | 0.32 | 0.198 | 0.06 | 1.82 | |
| ii. Full 50 s | Social-50 s | 0.96 | 0.013 | 0.92 | 0.99 |
| Model LR = − 41.06, AIC = 0.873 | MCHAT-R | 19.83 | < 0.001 | 4.26 | 92.25 |
| constant | 0.17 | 0.04 | 0.03 | 0.92 | |
| B. Sensitivity of GP-MCHA-R model at approx 90%, 80% and 70% specificity | Model AUC | Specificity approx 90% |
Specificity approx 80% |
Specificity approx 70% |
|||
|---|---|---|---|---|---|---|---|
| spec | sens | spec | sens | spec | sens | ||
|
Visit 1 Visit 2 |
0.89 0.88 |
92% 90% |
64% 63% |
78% 77% |
75% 81% |
69% 70% |
93% 85% |