FIGURE 2 from Integrative Prognostic Machine Learning Models in Mantle Cell Lymphoma

<p>Hyperparameter fitting and model metrics. <b>A,</b> A scatterplot showing the corresponding mean AUC (from cross-fold validation sets) to hyperparameter values tuned by the 50-grid Latin hypercube on the “all features” XGBoost model. <b>B,</b> Comparison of ROC curves from XGBoost Models. Of the trained models, the “all features” model performed the best on the test set, followed by the two “parsimonious” models (the iMIPI and iMIPI-s) and the combined “clinical and NGS” and “clinical and cytogenetic models.” <b>C,</b> Comparison of ROC curves from the “all features” and “parsimonious” (20-feature) XGBoost models and the multivariate linear model (GLM).</p>