Table 3 Classification outcomes using a 50% cut-off probability
A, Classification based on clinical characteristics alone | ||||
Training sample | ||||
 Predicted class | Survival <18 months | Survival >18 months | Sensitivity | 62.50% |
 True Class Survival <18 months | 15 | 7 | Specificity | 70.84% |
 Survival >18 months | 9 | 17 | Accuracy | 66.67% |
Validation sample | ||||
 Predicted class | Survival <18 months | Survival >18 months | Sensitivity | 66.67% |
 True Class Survival <18 months | 4 | 1 | Specificity | 83.34% |
 Survival >18 months | 2 | 5 | Accuracy | 75.00% |
B, Classification based on MRI measures alone | ||||
 Training sample | ||||
 Predicted class | Survival <18 months | Survival >18 months | Sensitivity | 79.16% |
 True Class Survival <18 months | 19 | 6 | Specificity | 75.00% |
 Survival >18 months | 5 | 18 | Accuracy | 77.08% |
Validation sample | ||||
 Predicted class | Survival <18 months | Survival >18 months | Sensitivity | 66.70% |
True Class Survival <18Â months | 4 | 3 | Specificity | 50.00% |
 Survival >18 months | 2 | 3 | Accuracy | 58.33% |
C, Classification based on clinical and MRI measures | ||||
Training sample | ||||
 Predicted class | Survival <18 months | Survival >18 months | Sensitivity | 75.00% |
 True Class Survival <18 months | 18 | 4 | Specificity | 83.34% |
 Survival >18 months | 6 | 20 | Accuracy | 79.17% |
Validation sample | ||||
 Predicted class | Survival <18 months | Survival >18 months | Sensitivity | 66.67% |
True ClassSurvival <18Â months | 4 | 1 | Specificity | 83.34% |
Survival >18Â months | 2 | 5 | Accuracy | 75.00% |