Skip to main content
Fig. 5 | BMC Cancer

Fig. 5

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. 5

Performance of the predictions for ccRCC patients, test on two different datasets. In gray without pre-screening, in blue after optimal bi-dimensional pre-screening performed on the TCGA dataset. In both case, the model is tested on an independent set not used for learning, either TCGA or ICGC. (A) C-index metric, higher is better. (B) IBS metric, lower is better. In each case, we computed the C-index by 10 repetitions of a nested 5-fold cross validation (section Suggested screening algorithm). Paired Wilcoxon tests. ***: \(p \le 0.001\), **: \(p \le 0.01\), *: \(p \le 0.05\), +: \(p \le 0.1\), n.s. : \(p > 0.1\)

Back to article page

BMC Cancer

ISSN: 1471-2407

Contact us