Skip to main content
. 2022 Dec 5;12:20975. doi: 10.1038/s41598-022-25389-7

Figure 1.

Figure 1

Comparison of average ROC curves across bootstrapped repetitions when clinical, radiomic, or clinical and radiomic features were used. The AUC for each curve is provided, along with the corrected AUC0.632+ value. The optimal upper-left operating point determined from the out-of-bag dataset is plotted as “+”, for which the MCR, FNR, and FPR were found. A false negative represents a BM that was predicted to respond to SRS, but instead progressed after SRS. 95% confidence intervals are provided for the curves (shaded bands) and error metrics (in parentheses).