Table 3 Performance of SVM prediction models
Models | AUC (95% CI) | Sensitivity (%) | Specificity (%) | Accuracy (%) |
|---|---|---|---|---|
Model based on conventional indicators | ||||
SSRIs improver - ADM non-improver | 0.776(0.684, 0.853) | 64.3(57.5, 71.9) | 79.5(69.8, 82.0) | 65.3(60.5, 76.8) |
SNRIs improver - ADM non-improver | 0.796(0.688, 0.882) | 72.4(60.1, 79.9) | 85.0(70.2, 89.7) | 73.0(62.5, 86.1) |
ADM improver -ADM non-improver | 0.799(0.674, 0.872) | 74.5(62.5, 83.2) | 81.3(69.0, 88.4) | 79.3(70.3, 89.4) |
Model based on radiomics features | ||||
SSRIs improver - ADM non-improver | 0.954(0.912, 0.980) | 89.2(78.8, 95.2) | 87.4(79.4, 93.1) | 88.5(82.4, 92.5) |
SNRIs improver - ADM non-improver | 0.942(0.897, 0.971) | 91.9(83.2, 96.6) | 82.5(73.8, 89.3) | 86.8(80.5, 91.1) |
ADM improver -ADM non-improver | 0.889(0.798, 0.952) | 91.2(76.8, 98.1) | 80.1(64.4, 90.9) | 85.1(75.0, 92.3) |