A Genomic-Clinicopathologic Nomogram for the Prediction of Lymph Node Invasion in Prostate Cancer

Figure 2

Construction of SVM models. (a) Gene selection process using SVM-RFE and 10-fold cross-validation in the imbalanced training set: 16 genes with the highest discriminative accuracy were selected for SVM model development. (b) SVM-RFE is used to rank genes according to the gene importance, and the top 16 genes were selected for SVM model development. (c) Gene selection process using SVM-RFE and 10-fold cross-validation in the SMOTE-balanced training set: 37 genes with the highest discriminative accuracy were selected for SVM model development. (d) SVM-RFE is used to rank genes according to the gene importance, and the top 37 genes were selected for SVM model development. SVM: support vector machine; SVM-RFE: SVM-based recursive feature elimination; SMOTE: synthetic minority oversampling technique.

(a)

(b)

(c)

(d)