Fig. 4

Risk prediction done for the cohort of 1273 patients of CRC based in the multivariate analysis using the top 100 genes that showed up-regulation correlated with poor prognosis (i.e. overexpressed in low survival cases). a Plot presenting the patients according to their risk score, from Low (blue) to High (red) risk. A recursive algorithm using 10-fold cross-validation finds the value of risk score (marked with a vertical black line) that allows the best splitting of the cohort in two groups. b Kaplan-Meier plot showing the separation of these two groups: a high-risk group including 425 individuals (in red) and a low-risk group including 848 individuals (in blue). The analysis has been done using a multivariate Cox proportional-hazards regression. As shown, the division is very significant (p-value = 8.25e-14) and allows an optimal separation of individuals according to their survival. The analysis of the beta factors assigned by the regression to each of the top 100 genes (i.e. to each variable within the multivariate vector) allows the identification of the genes that are the most influential factors in this risk analysis and therefore it helps in the selection of the best “gene survival markers”