Table 2 Predictive performance of radiomics model, clinical model, and radiomics–clinical model
Model | Training (N = 122) | Internal test (N = 46) | External test (N = 55) | ||||||
|---|---|---|---|---|---|---|---|---|---|
Radiomics | Clinical | Combination | Radiomics | Clinical | Combination | Radiomics | Clinical | Combination | |
AUC (95% CI) | 0.835 (0.76, 0.90) | 0.705 (0.61, 0.79) | 0.865 (0.79, 0.92) | 0.806 (0.67, 0.93) | 0.714 (0.56, 0.86) | 0.824 (0.69 0.95) | 0.765 (0.63, 0.89) | 0.699 (0.55, 0.84) | 0.773 (0.64, 0.90) |
Sensitivity | 71.2% | 66.1% | 76.6% | 76.2% | 66.7% | 80.0% | 72.4% | 73.1% | 75.0% |
Specificity | 76.8% | 68.3% | 81.3% | 76.0% | 64.3% | 76.9% | 61.5% | 62.1% | 66.7% |
Accuracy | 73.8% | 67.2% | 78.7% | 76.1% | 65.2% | 78.3% | 67.3% | 67.3% | 70.9% |