Research Paper Volume 12, Issue 21 pp 21481—21503

Two machine learning methods identify a metastasis-related prognostic model that predicts overall survival in medulloblastoma patients

Figure 1. Comparison of the predictive power of the 47-mRNA metastasis-related model based on Lasso-penalized Cox regression and the 21-mRNA metastasis-related model based on random survival forest-variable hunting (RSF-VH). The receiver operating characteristic (ROC) curves of the 47-mRNA-based model based on Lasso-penalized Cox regression (A) and the 21-mRNA-based model based on RSF-VH (B).