Fig. 9

(A) Feature selection using LASSO. where the non-zero coefficients of 7 mRNA and lncRNA as features was calculated and displayed. B The ROC curve showed the prediction accuracy of different six models on the test set. The AUC value of each model represents the area under the ROC curve. The closer the AUC to 1, the better the generalization of the model. C The back-propagation neural network (BPNN) provided the most efficient estimates. D RT-qPCR analysis of LINC00974 expression in 10 CRC and adjacent normal tissue samples