| Â | Â | Â | Ratioscore | SVM |
|---|
| Â | Control | Case | Sensitivity | Specificity | Sensitivity | Specificity |
|---|
#1 | CONTROL | MS | 0.87 | 1.00 | 0.87 | 0.97 |
CONTROL | CIS âž” MS | 0.96 | Â | 0.95 | Â |
#2 | OND | MS | 0.70 | 1.00 | 0.82 | 0.78 |
OND | CIS âž” MS | 1.00 | Â | 1.00 | Â |
#3 | OND-NI | MS | 0.86 | 1.00 | 0.84 | 0.94 |
OND-NI | CIS âž” MS | 1.00 | Â | 1.00 | Â |
#4 | OND-I | MS | 0.90 | 1.00 | 0.77 | 0.93 |
OND-I | CIS âž” MS | 1.00 | Â | 0.98 | Â |
- Optimum ratios for the ratioscore method were from Figures 2, 3 and Additional file 3: Figure S3. CIS ➔ MS subject data were inputted and scores computed. For the SVM, 60% of controls and cases were randomly selected for the training set and 40% were used for the validation set. Sensitivity and specificity were calculated for the combined sets. These results defined the SVM. CIS ➔ MS subject data were applied to the SVM and subjects received a score of 0 if assigned to the CONTROL cohort or 1 if assigned to the CASE cohort. Sensitivity was calculated from this output.
- Sensitivity = # true positives/(# true positives + # false negatives).
- Specificity = # true negatives/(# true negatives + # false positives).
