Skip to main content
Fig. 3 | Cardiovascular Diabetology

Fig. 3

From: A novel 6-metabolite signature for prediction of clinical outcomes in type 2 diabetic patients undergoing percutaneous coronary intervention

Fig. 3

The variable selection process for identifying a 6-metabolite signature to predict 1-year MACEs after PCI among patients with T2DM. A Hierarchical cluster analysis of the 35 metabolites differentially expressed in both discovery and internal validation sets. B Of the 35 metabolites, 27 important metabolites with some redundancy are selected in the first step of a variable selection algorithm (VSURF). C Of the remaining 27 metabolites, 11 metabolites that avoid redundancy are further selected in the second step of VSURF. D The final step of VSURF identifies a combination of 6 metabolites with the best potential for predicting 1-year MACEs. The OPLS-DA scatter plot based on the 6 metabolites in both discovery (E) and internal validation (F) sets. G The ROC analysis of the 6 metabolites based on logistic regression. ADPR adenosine diphosphate ribose; 1-MNAM 1-Methylnicotinamide; PC phosphatidylcholine; NAM nicotinamide

Back to article page