Table 2 Prediction of remission for bDMARDs
Follow-up period (year), mean (sd) | Remission/total | Measurement | Baseline | Lasso | Ridge | SVM | Random forest | Xgboost | |
|---|---|---|---|---|---|---|---|---|---|
All bDMARDs | 0.96 (0.30) | 564/1397 (40.4%) | Sensitivity | 0.0% | 33.1% | 29.6% | 24.9% | 0.6% | 32.2% |
Specificity | 100.0% | 79.9% | 83.1% | 85.1% | 100.0% | 80.5% | |||
Accuracy | 59.6% | 61.0% | 61.5% | 60.5% | 59.8% | 62.0% | |||
AUROC | 0.500 | 0.614 | 0.619 | 0.602 | 0.608 | 0.596 | |||
TNF inhibitors | 0.93 (0.32) | 252/793 (31.8%) | Sensitivity | 0.0% | 10.7% | 21.3% | 9.3% | 0.0% | 20.0% |
Specificity | 100.0% | 97.5% | 92.6% | 96.0% | 100.0% | 91.4% | |||
Accuracy | 68.2% | 69.6% | 70.0% | 68.4% | 68.4% | 68.8% | |||
AUROC | 0.500 | 0.649 | 0.655 | 0.633 | 0.644 | 0.637 | |||
Non-TNF inhibitors | 1.01 (0.27) | 312/604 (51.7%) | Sensitivity | 0.0% | 62.4% | 64.5% | 62.4% | 75.3% | 57.0% |
Specificity | 100.0% | 52.3% | 51.7% | 51.7% | 33.3% | 48.9% | |||
Accuracy | 51.7% | 57.8% | 57.8% | 58.3% | 55.3% | 52.8% | |||
AUROC | 0.500 | 0.605 | 0.607 | 0.606 | 0.586 | 0.538 | |||
Adalimumab | 0.93 (0.31) | 91/289 (31.5%) | Sensitivity | 0.0% | 22.2% | 29.6% | 37.0% | 0.0% | 14.8% |
Specificity | 100.0% | 91.5% | 88.1% | 81.4% | 100.0% | 93.2% | |||
Accuracy | 68.5% | 69.8% | 69.8% | 67.4% | 68.6% | 69.8% | |||
AUROC | 0.500 | 0.680 | 0.688 | 0.663 | 0.623 | 0.629 | |||
Etanercept | 0.99 (0.34) | 75/220 (34.1%) | Sensitivity | 0.0% | 29.5% | 36.4% | 40.9% | 0.0% | 22.7% |
Specificity | 100.0% | 86.0% | 83.7% | 79.1% | 100.0% | 88.4% | |||
Accuracy | 65.9% | 67.7% | 67.7% | 66.2% | 66.2% | 64.6% | |||
AUROC | 0.500 | 0.634 | 0.656 | 0.643 | 0.656 | 0.619 | |||
Golimumab | 0.97 (0.30) | 41/122 (33.6%) | Sensitivity | 0.0% | 0.0% | 41.7% | 50.0% | 0.0% | 16.7% |
Specificity | 100.0% | 100.0% | 79.2% | 66.7% | 100.0% | 95.8% | |||
Accuracy | 66.4% | 66.7% | 63.9% | 61.1% | 66.7% | 66.7% | |||
AUROC | 0.500 | 0.615 | 0.694 | 0.623 | 0.659 | 0.635 | |||
Infliximab | 0.85 (0.32) | 45/162 (27.8%) | Sensitivity | 0.0% | 30.8% | 23.1% | 30.8% | 0.0% | 15.4% |
Specificity | 100.0% | 82.9% | 88.6% | 80.0% | 100.0% | 85.7% | |||
Accuracy | 72.2% | 66.7% | 70.8% | 64.6% | 72.9% | 66.7% | |||
AUROC | 0.500 | 0.595 | 0.626 | 0.544 | 0.600 | 0.511 | |||
Abatacept | 0.99 (0.30) | 62/194 (32.0%) | Sensitivity | 0.0% | 11.1% | 30.6% | 38.9% | 0.0% | 27.8% |
Specificity | 100.0% | 94.9% | 84.6% | 75.6% | 100.0% | 82.1% | |||
Accuracy | 68.0% | 68.4% | 68.4% | 63.2% | 68.4% | 64.9% | |||
AUROC | 0.500 | 0.635 | 0.679 | 0.636 | 0.618 | 0.598 | |||
Tocilizumab | 1.01 (0.26) | 250/410 (61.0%) | Sensitivity | 0.0% | 76.0% | 81.3% | 80.0% | 80.0% | 77.3% |
Specificity | 100.0% | 29.2% | 22.9% | 22.9% | 21.9% | 22.9% | |||
Accuracy | 61.0% | 57.7% | 58.5% | 61.0% | 56.9% | 56.1% | |||
AUROC | 0.500 | 0.552 | 0.555 | 0.556 | 0.522 | 0.512 |