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

Figure 6

Performance of the SVM-based risk score in the SMOTE-balanced dataset. Alluvial diagrams showed the predictive results of the 37-gene-based SVM model in SMOTE-balanced training (a) and validation (b) sets. Violin plots of the risk score in SMOTE-balanced training (c) and validation (d) sets grouped by the N stage. Kaplan–Meier curves of DFS for prostate cancer patients based on the risk group in SMOTE-balanced training (e) and validation (f) sets. DFS: disease-free survival; FP: false positive; FN: false negative; SVM: support vector machine; SMOTE: synthetic minority oversampling technique.

(a)

(b)

(c)

(d)

(e)

(f)