Table 3.
Comparison of the performance of the models measured by difference in the area-under-the-curve (AUCs) across 1000 permuted runs. Each entry in the table has p-value, average difference of AUC of the models, and confidence interval of 1000 AUCs.
| Classification models | Ia: Imaging | Ib: Texture | Ic: Clinical | IIa: Imaging+Texture | IIb: Imaging+Clinical | IIc: Texture+Clinical |
|---|---|---|---|---|---|---|
| Ia:Imaging | --- | 0.530, −0.01 (−0.27, 0.25) |
0.636, −0.05 (−0.34,0.25) | --- | --- | --- |
| Ib:Texture | --- | --- | 0.598, −0.05 (−0.31,0.25) | --- | --- | --- |
| IIa:Imaging+Texture | 0.045, 0.09 (−0.08,0.34) | 0.048, 0.10 (−0.06,0.35) | 0.032, 0.20 (−0.06,0.42) | --- | 0.606, −0.02 (−0.19,0.15) | 0.562, −0.01 (−0.17,0.15) |
| IIb:Imaging+Clinical | 0.173, 0.12 (−0.11,0.34) | 0.123, 0.13 (−0.09,0.34) | 0.109, 0.17 (−0.09,0.41) | --- | --- | 0.457, 0.01 (−0.17,0.19) |
| IIc:Texture+Clinical | 0.146, 0.13 (−0.09,0.34) | 0.117, 0.14 (−0.09,0.33) | 0.102, 0.18 (−0.08,0.41) | --- | --- | --- |
| IIIa:Imaging+Texture+Clinical | 0.002, 0.26 (0.07,0.43) | 0.001, 0.27 (0.09,0.44) | 0.001, 0.32 (0.08,0.50) | 0.004, 0.12 (0.01,0.23) | 0.009, 0.14 (0.02,0.27) | 0.010, 0.13 (0.02,0.27) |