Fig. 5
Screening of prognostic MRLs. A LASSO regression of the eight optimal MRLs. B Cross-validation for tuning the parameter selection and showing confidence interval under each lambda the in the LASSO regression. The two dashed lines indicate two special λ values: λmin on the left and λ1se on the right. The λ values between these two values were considered to be appropriate. The model constructed by λ1se was the simplest, that was, it used a small number of genes, while λmin had a higher accuracy rate and used a larger number of genes. Hence, λmin was selected to build the model for accuracy in our study. C The forest map showing the results of Cox regression analysis and LASSO regression coefficient of the eight optimal MRLs. The values in parentheses represented 95% confidence intervals. D-K Survival analysis of the eight optimal MRLs for RP5-1120P11.1 (D), RP11-195F19.9 (E), USP30-AS1 (F), AC004540.5 (G), ZFAS1 (H), RP11-10A14.5 (I), AC010761.10 (J), and AC003075.4 (K). The P values are tested by log-rank