Table 4.
Cut-off point identified by ROC analysis∗1.
| Number of dichotomized variables∗2 |
Sensitivity | Specificity | Sensitivity + specificity |
Correctly classified | LR+ | LR− |
|---|---|---|---|---|---|---|
| 0 | 1.000 | 0.000 | 1.000 | 0.134 | 1.000 | — |
| 1 | 1.000 | 0.027 | 1.027 | 0.158 | 1.028 | 0.000 |
| 2 | 1.000 | 0.073 | 1.073 | 0.197 | 1.078 | 0.000 |
| 3 | 1.000 | 0.182 | 1.182 | 0.291 | 1.222 | 0.000 |
| 4 | 0.941 | 0.332 | 1.273 | 0.413 | 1.409 | 0.177 |
| 5 | 0.912 | 0.468 | 1.380 | 0.528 | 1.714 | 0.189 |
| 6 | 0.794 | 0.700 | 1.494 | 0.713 | 2.647 | 0.294 |
| 7 | 0.618 | 0.868 | 1.486 | 0.835 | 4.686 | 0.440 |
| 8 | 0.382 | 0.941 | 1.323 | 0.866 | 6.471 | 0.656 |
| 9 | 0.177 | 0.986 | 1.163 | 0.878 | 12.941 | 0.835 |
| 10 | 0.029 | 1.000 | 1.029 | 0.870 | — | 0.971 |
LR+: likelihood ratio for a positive test result; LR−: likelihood ratio for a negative test result.
∗1Dataset was retrieved from our previous study [9].
∗2The number of dichotomized variables was the cumulated effects of the various clinicopathologic variables from Table 3, including tumor size, stage, ER, HT, LN metastasis, age, PR, BMI, MRE11 positive cells, and HER2.