Wnt/β-Catenin, Carbohydrate Metabolism, and PI3K-Akt Signaling Pathway-Related Genes as Potential Cancer Predictors

Figure 1

Average areas under the curve (AUCs) for Wnt signal pathway-related genes and differentially expressed genes (DEGs) using four machine learning algorithms to predict colorectal cancer from gene expression data. For the pathway genes, support-vector machine (SVM) yields an AUC of 99.49%, decision tree (DT) yields 89.45%, random forest (RF) yields 99.49%, and k-nearest neighbor (KNN) yields 99.42%. For DEGs, SVM yields 99.49%, DT, 99.49%, RF, 96.18%, and KNN, 97.85%.