Hello everyone,

I am struggling with generating the ROC curve for the DIABLO model for new (testing) data. For context I created a DIABLO model with 3 omics blocks and 4 components. Even obtaining continuous predictions instead of the binary ones would be very useful for my work.

This is the part of my code where I was trying to get the continuous predictions:

# Generate predictions for the test set using the joint model
predictions <- predict(final.diablo.model, newdata = X.test, dist = "mahalanobis.dist")

# Get the continuous prediction scores for AUC calculation
pred_cont_met <- predictions$variates$Metabolomics[, ncomp]
pred_cont_lip <- predictions$variates$Lipidomics[, ncomp]
pred_cont_prot <- predictions$variates$Proteomics[, ncomp]

Thank you a lot in advance :slight_smile:

hi @majagojska,

There was a. post about AUC just recently, have a look. We already plot AUROC and you can extract this already.