Classification and survival analysis using N-MEG genes. (A) An ROC curve shows that the genes from MEGA analysis (N-MEG) have the strongest classification power among other candidate lymph node related genes. (B) Areas under ROC curve (AUC) denote the statistical measurement of the classification power. Note that the genes from previous two-class comparisons hardly prove their usefulness in classification; but the result may differ when conducted in other classification methods. (C and D) Three risk groups (good, intermediate and poor) show distinct survival and hazard functions in Kaplan-Meyer diagrams. AUC=area under curve, Std. Error=standard error, Asym. Prob.=Asymptotic probability, LCL=lower confidence limit, UCL=upper confidence limit.