Fig. 4

This study utilized machine learning to identify feature genes associated with COPD. (A) Presents Venn diagrams comparing differential genes between Control and COPD, genes from the WGCNA module most correlated with COPD, and senescence-related genes. (B,C) Describe the use of the support vector machine recursive feature elimination (SVM-RFE) algorithm to select biometric feature genes, identifying the optimal number based on the lowest accuracy and error rate. (D) Details the application of the LASSO algorithm for feature gene selection. (E) Examines the relationship between the number of decision trees and the error rate in the Random Forest algorithm. (F) Highlights the top six genes identified by Random Forest in terms of gene importance. Finally, (G) showcases a Venn diagram summarizing feature genes identified by all three machine learning algorithms as key markers for COPD.