Model * | No. of variables | AUC(95%CI) | Sensitivity | Specificity | Youden index | Cut off value | P for H–L test # |
---|
Model 1 | 1 | 0.83(0.77–0.89) | 0.820 | 0.683 | 0.504 | 0.621 | 0.923 |
Model 2 | 2 | 0.86(0.80–0.91) | 0.789 | 0.833 | 0.622 | 0.640 | 0.390 |
Model 3 | 3 | 0.87(0.82–0.92)a | 0.781 | 0.833 | 0.615 | 0.653 | 0.688 |
Model 4 | 4 | 0.88(0.84–0.93)ab | 0.867 | 0.767 | 0.634 | 0.569 | 0.289 |
Model 5 | 5 | 0.90(0.85–0.94)ab | 0.813 | 0.850 | 0.663 | 0.698 | 0.873 |
Model 6 | 6 | 0.91(0.86–0.95)abcd | 0.695 | 0.950 | 0.645 | 0.828 | 0.148 |
Model 7 | 7 | 0.91(0.87–0.95)abcd | 0.766 | 0.900 | 0.666 | 0.762 | 0.632 |
Model 8 | 8 | 0.92(0.89–0.96)abcd | 0.922 | 0.800 | 0.722 | 0.494 | 0.049 |
- *Model 1: logit(p1) = -2.365 + 0.572 × "PSQI score"; Model 2: logit(p2) = -2.954 + 0.547 × "PSQI score" + 0.034 × "Qi-depressed score"; Model 3: logit(p3) = -2.606 + 0.599 × "PSQI score" + 0.046 × "Qi-depressed score"-0.039 × "Inherited special score"; Model 4: logit(p4) = -2.667 + 0.581 × "PSQI score" + 0.039 × "Qi-depressed score" + 0.034 × "Blood-stasis score"-0.05 × "Phlegm-dampness score"; Model 5: logit(p5) = -2.267 + 0.554 × "PSQI score" + 0.046 × "Qi-depressed score" + 0.041 × "Qi-deficient score"-0.04 × "Phlegm-dampness score"-0.043 × "Inherited special score"; Model 6: logit(p6) = -2.446 + 0.574 × "PSQI score" + 0.036 × "Qi-depressed score" + 0.039 × "Qi-deficient score" + 0.048 × "Blood-stasis score"-0.059 × "Phlegm-dampness score"-0.055 × "Inherited special score"; Model 7: logit(p7) = -0.84 + 0.612 × "PSQI score"-0.057 × "SDS score" + 0.046 × "Qi-depressed score" + 0.052 × "Qi-deficient score" + 0.051 × "Blood-stasis score"-0.063 × "Phlegm-dampness score"-0.057 × "Inherited special score"; Model 8: logit(p8) = -2.687 + 0.532 × "PSQI score" + 0.12 × "SAS score"-0.111 × "SDS score" + 0.043 × "Qi-depressed score" + 0.055 × "Qi-deficient score" + 0.051 × "Blood-stasis score"-0.068 × "Phlegm-dampness score"-0.058 × "Inherited special score"
- #indicated the P value for the Hosmer–Lemeshow test for various model
- aindicated that after Delong test, the AUC was significantly better than model 1
- bindicated that after Delong test, the AUC was significantly better than model 2
- cindicated that after Delong test, the AUC was significantly better than model 3
- dindicated that after Delong test, the AUC was significantly better than model 4
- AUC Area under the receiver operating characteristic curve, H–L Hosmer–Lemeshow