Model Evaluation Dashboard

Move a single decision threshold and watch it reshape the ROC curve, the calibration plot, the probability distribution and the decision curve at once.

ROC
calibration
net benefit
threshold
external validation

AUC

0.000

Sensitivity

0.000

Specificity

0.987

Net benefit @ pt

-0.037

ROC curve

Precision-recall curve

Calibration plot

Decision curve

ModelTreat allTreat none

Reading the dashboard

Discrimination, calibration and utility are separate questions answered by separate plots. A model can have AUC below 0.5 on one cohort and still, at an extreme threshold, edge out treat-none — but that does not make it clinically useful. Compare the model curve against both defaults, not just against zero.

Preparing the simulation…