Table 4.
Results of ROC analysis (AIS versus MIA versus IPA).
| Training set |
Validation set |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| AUC (95% CI) | P | Cut-off | Sen. (%) | Spec. (%) | AUC (95% CI) | P | Cut-off | Sen. (%) | Spec. (%) | |
| AIS vs IPA | ||||||||||
| MTD | 0·855 (0·736–0·974) | <001 | 1·3 | 72·7 | 88·2 | 0·831 (0·650–1·012) | 009 | 1·3 | 63·6 | 71·4 |
| FR+-CTC | 0·757 (0·627–0·887) | 002 | 9·2 | 75·0 | 70·6 | 0·727 (0·492–0·963) | 08 | 9·2 | 68·2 | 57·1 |
| Combined | 0·893 (0·795–0·991) | <001 | 0·767 | 81·8 | 88·2 | 0·864 (0·676–1·052) | 004 | 0·767 | 72·7 | 85·7 |
| AIS vs MIA + IPA | ||||||||||
| MTD | 0·763 (0·631–0·896) | <001 | 1·1 | 66·7 | 76·5 | 0·758 (0·562–0·955) | 03 | 1·1 | 67·7 | 71·4 |
| FR+-CTC | 0·716 (0·593–0·840) | 006 | 10·3 | 53·0 | 88·2 | 0·685 (0·457–0·913) | 13 | 10·3 | 55·9 | 85·7 |
| Combined | 0·804 (0·687–0·921) | <001 | 0·805 | 69·7 | 82·4 | 0·824 (0·604–1·043) | 008 | 0·805 | 67·7 | 71·4 |
| AIS + MIA vs IPA | ||||||||||
| MTD | 0·863 (0·783–0·942) | <001 | 1·4 | 68·2 | 92·3 | 0·787 (0·646–0·928) | 002 | 1·4 | 63·6 | 79·0 |
| FR+-CTC | 0·650 (0·529–0·771) | 02 | 8·8 | 84·1 | 46·2 | 0.673 (0·508–0·839) | 06 | 8·8 | 77·3 | 42·1 |
| Combined | 0·878 (0·803–0·954) | <001 | 0·577 | 75·0 | 89·7 | 0·830 (0·707–0·953) | <001 | 0·577 | 63·6 | 73·7 |
Only patients who were randomly selected to receive serum biomarker tests (17 AIS, 22 MIA, and 44 IPA patients in the training set; 7 AIS, 12 MIA, and 22 IPA patients in the validation set) were included in this ROC analysis.