Skip to main content
Fig. 2 | BMC Cancer

Fig. 2

From: Prognosis of lasso-like penalized Cox models with tumor profiling improves prediction over clinical data alone and benefits from bi-dimensional pre-screening

Fig. 2

C-indices obtained with clinical data alone (red), mRNA-seq data alone with ridge penalty and all the genes (blue, no bi-dimensional pre-screening), and clinical and mRNA-seq data together (purple) for the 16 cancers studied. We computed the C-indices by 10 repetitions of a K-fold cross-validation (K=5). To test whether the added value of mRNA-seq data for prediction is significant over clinical data alone, we computed the p-values of a one-sided Wilcoxon signed-rank test (purple stars at the top of each graphic, Benjamini-Hochberg correction for the 16 p-values). Red letters at the bottom of each graphics indicate the clinical data available (G: grade; T: tumor; N: node; M: metastasis). Age is available for all cancers, and gender only for non-unisexual cancers (CESC, PRAD, TGCT are sex-specific). ***: \(p \le 0.001\), **: \(p \le 0.01\), *: \(p \le 0.05\), +: \(p \le 0.1\), n.s. : \(p > 0.1\)

Back to article page