Table 2.
Performance comparison between our method and the three state-of-the-art prediction methods
| Methods | sample size | cv type | Specificity | Sensitivity | Accuracy | AUC |
|---|---|---|---|---|---|---|
| RLSMDA | 1184+ | LOOCV | – | – | – | 0.9475 |
| our model | 1184+,1184- | LOOCV | 0.9367 | 0.9368 | 0.9367 | 0.9896 |
| Xu’s method | 37+, 44- | 5-fold | 0.8833 | 0.8643 | 0.8772 | 0.9189 |
| our model | 37+, 37- | 5-fold | 0.9990 | 1.000 | 0.9995 | 0.9854 |
| Jiang’s method | 270+, 270- | 10-fold | 0.9125 | 0.7338 | 0.8232 | 0.8884 |
| our model | 263+, 263- | 10-fold | 0.9274 | 0.8982 | 0.9128 | 0.9871 |
Symbols “+/-” represent “positive samples/negative samples”. cv means cross-validation
The best performance among the compared methods are showed in boldface