Machine Learning Predicts the Oxidative Stress Subtypes Provide an Innovative Insight into Colorectal Cancer

Figure 3

8-gene signature model to predict CRC patients’ outcome. DEGs between C1 and C2 in total datasets (a), Lasso regression discover 15 genes that are important (b&c), after multivariate Cox and eight model genes are selected and eight genes signature are built (d). In this model, patients with low-risk score always mean a better outcome, when compared with high-risk score patients (e), and the ROC demonstrates that this model has a good predictive ability for patients’ prognosis (f).

(a)

(b)

(c)

(d)

(e)

(f)