ROC Analysis Options :
Classical univariate ROC curve analyses
Perform classical univariate ROC curve analysis, such as to generate ROC curve, to calculate AUC or partial AUC as well as their 95% confidence intervals, to compute optimal cutoffs for any given feature, as well as to generate performance tables for sensitivity, specificity, and confidence intervals at different cutoffs.
Multivariate ROC curve based exploratory analysis (Explorer)
Perform automated important feature identification and performance evaluation. ROC curve analyses are performed based on three multivariate algorithms - support vector machines (SVM), partial least squares discriminant analysis (PLS-DA), and random forests.
ROC curve based model evaluation (Tester)
Users can manually select any combination of features to create biomarker models using any of the three algorithms mentioned above. The module also allows users to hold out a subset of samples for extra validation purpose, as well as to predict class for new samples (samples without class labels).