Table 4.
Missing baseline data | Model name | Training data (n) | CV/test | AUC | Overall accuracy (%) | Sens (%) |
Spec (%) |
OOP |
---|---|---|---|---|---|---|---|---|
None | CG | 13 936 | CV | 0.89 | 81 | 77 | 83 | 0.42 |
test | 0.89 | 81 | 75 | 85 | 0.42 | |||
Genotype | CNG | 56 717 | CV | 0.84 | 77 | 71 | 80 | 0.43 |
test | 0.84 | 76 | 71 | 79 | 0.43 | |||
Time on therapy | CG(−ToT) | 19 080 | CV | 0.90 | 82 | 80 | 84 | 0.42 |
test | 0.90 | 82 | 80 | 83 | 0.42 | |||
Genotype/time on therapy | CNG(−ToT) | 56 717 | CV | 0.83 | 76 | 71 | 80 | 0.44 |
test | 0.82 | 76 | 71 | 79 | 0.44 | |||
Genotype/CD4 count | CNG(−CD4) | 62 940 | CV | 0.83 | 76 | 72 | 79 | 0.44 |
test | 0.82 | 75 | 73 | 77 | 0.44 | |||
Genotype/time on therapy/ CD4 count | CNG(−ToT −CD4) | 50 270 | CV | 0.79 | 72 | 71 | 73 | 0.38 |
test | 0.78 | 72 | 74 | 70 | 0.38 |
CV, cross-validation; test, independent testing; sens, sensitivity; spec, specificity.