Figure 2

The clinical impact curves and net reclassification improvement (NRI) and Receiver operating characteristic curve (ROC) of the diagnostic models in the training and external validating cohort. The clinical impact curves of the WLBD (a), WBD (b), and WD (c) were plotted to verify the clinical utility of these models in the training cohort. The NRI of WD between WBD (d) was calculated to identify the optimization model. The ROC curve (e) and calibration plots (f) of WBD model were plotted in the external validating cohort.