Skip to main content
Fig. 4 | BMC Cancer

Fig. 4

From: LUADpp: an effective prediction model on prognosis of lung adenocarcinomas based on somatic mutational features

Fig. 4

Inter-stage prediction of LUAD prognosis with the genetic models based on somatic mutation features. a. Overlap of the top genes with mutation rate difference between good and poor sub-groups for patients diagnosed at all, early and later stages. The name was shown for the gene shared by all the three stratifications. b. The ROC curves of SVM models based on different sizes of somatic mutation features. The curves were based on the prediction results of 5-fold cross-validation testing datasets. c. Performance comparison of different genetic models. The 5-fold cross-validation results were compared for ROC_AUC and Accuracy, and the standard deviations were shown in error bars. Students’ pairwise t tests were performed. d. The inter-stage predictive ROC curves of genetic models. The ‘early -> later’ represented prediction of later patients with the model trained with early cases while the ‘later -> early’ represented prediction of early patients with the model trained with later cases. e. Survival curves of sub-groups of cases classified with different inter-stage models

Back to article page