Model | â–³AUC | AUC | SEN | SPC | PPV | NPV | Accuracy |
---|
Model 4Aa | â–³A-C | 0.88 | 0.83 | 0.80 | 0.81 | 0.83 | 0.82 |
Model 4Bb | â–³A-B* | 0.79 | 0.82 | 0.76 | 0.77 | 0.81 | 0.79 |
Model 4Cc | â–³B-C* | 0.88 | 0.82 | 0.82 | 0.82 | 0.82 | 0.82 |
- AUC area under curve, SEN sensitivity, SPC specificity, PPV positive predictive value, NPV negative predictive value
- △AUC = difference in AUC between various models
- â–³A-B*: significant difference in AUC between Model 4A and Model 4B
- (p < 0.00001); △B-C*: the same meaning as that of △A-B* (p < 0.00001); △A-C: no significant difference in AUC between Model 4A and Model 4C (p = 0.785)
- aModel 4A
- Propensity model:
- Y = 0.963*BSRS-5 item 2 + 0.564*BSRS-5 item 4–0.655*BSRS-5 item 5 + 6.285
- propensity score = expY/(1 + expY);
- For a given individual, if the propensity score > 0.0355, then the model was predictive of suicidal ideation
- bModel 4B
- BSRS-5 total score greater than 5 coded as 1, otherwise as 0
- Propensity model:
- Y = 2.72*(BSRS-5 total score greater than 5)-4.911
- propensity score = expY/(1 + expY);
- For a given individual, if the propensity score > 0.0539, then the model was predictive of suicidal ideation
- cModel 4C
- Propensity model:
- Y = 1.226*BSRS-5 item 1 + 2.337*BSRS-5 item 2 + 1.297*BSRS-5 item 3
- + 1.646*BSRS-5 item 4–1.694*BSRS-5 item 5 + 6.374
- propensity score = expY/(1 + expY);
- For a given individual, if the propensity score > 0.0386, then the model was predictive of suicidal ideation