Skip to main content
Fig. 6 | BMC Cancer

Fig. 6

From: Retrospective validation study of an artificial neural network-based preoperative decision-support tool for noninvasive lymph node staging (NILS) in women with primary breast cancer (ISRCTN14341750)

Fig. 6

The entire cohort, batch-mode results using both preoperative and postoperative data. A) Area under the receiver operating characteristics curve (AUC) visualizing discriminatory performance of noninvasive lymph node staging (NILS) model for the estimation of axillary disease (N+). B) Hosmer–Lemeshow calibration plot of observed proportion N+ versus mean predicted probability of N+ for each decile of the predictions. Locally weighted scatterplot smoothing (LOWESS), the dotted line, was used to capture the calibration performance for low probabilities of N+ , i.e. within the red box

Back to article page