Fig. 3

Construction of OSCC prognostic model. A The forest plot of seven genes significantly associated with the prognosis of OSCC by univariate Cox regression analysis. HR is short for Hazard ratio and 95%CI is 95% of the confidence interval. B The graph to determine the best lambda by LASSO Cox model. The x-axis is the log (lambda) and y-axis is the partial likelihood deviance whose smallest value is the best lambda value. C The Kaplan–Meier survival curve in TCGA dataset. The x-axis is time and y-axis is the survival probability. The red is the low-risk group and the blue is the high-risk group. p value is calculated by the log-rank test. D The Kaplan–Meier survival curve in meta-GEO dataset. The x-axis is time and y-axis is the survival probability. The red is the low-risk group and the blue is the high-risk group. p value is calculated by the log-rank test. E and F The estimated Risk Score of each patient is ranked from small to big. The vertical dotted line is the median of Risk Score