Screening of Biomarkers for Early Diagnosis of Coronary Artery Disease and Prediction of Chinese Materia Medica Treatment by Integrating Bioinformatics and Machine Learning

YU Jiatian; ZHOU Conghui; TONG Xu

doi:10.3969/j.issn.1006-3250.2026.03.008

Abstract：Objective To integrate bioinformatics and machine learning methods to screen biomarkers for early diagnosis of coronary artery disease(CAD)and predict effective traditional Chinese medicine(TCM),providing a basis and methodological reference for the application of TCM in the early prevention and treatment of CAD.Methods This study obtained peripheral blood miRNA expression data of CAD patients from the GEO database and used R software to screen differentially expressed genes(DEGs).Weighted gene co-expression network analysis(WGCNA)was employed to identify co-expressed gene modules with high biological significance.Enrichment analysis was performed on the overlapping genes between DEGs and the modules screened by WGCNA.The LASSO algorithm was used to predict characteristic genes of CAD,and hub genes were screened based on the STRING database and protein-protein interaction(PPI)network.The overlapping genes were considered potential biomarkers.Subsequently,the differential expression of key genes was validated in different datasets,and their diagnostic value was evaluated using receiver operating characteristic(ROC)curves.Gene set enrichment analysis(GSEA)was conducted to identify pathways significantly enriched for key genes in CAD.Finally,based on the key genes,targeted TCM for CAD treatment was predicted using the Coremine Medical platform,and their efficacy,properties,and meridian tropism were analyzed in combination with TCM theory.Results A total of 391 DEGs were identified in this study,and 156 genes were screened through WGCNA.The intersection yielded 48 genes closely associated with CAD.Enrichment analysis indicated that these genes were significantly enriched in pathways such as the peroxisome proliferator-activated receptor(PPAR)signaling pathway and osteoclast differentiation.Through algorithms such as LASSO regression,aquaporin 9(AQP9)and FK-506 binding protein 5(FKBP5)were screened as potential biomarkers for CAD,and ROC curve analysis confirmed their high diagnostic efficacy(AUC>0.78).GSEA further revealed that AQP9 and FKBP5 were significantly enriched in biological processes such as immune regulation,metabolic regulation,and cellular signal transduction in CAD.Finally,15 types of Chinese materia medica,including Chishao(Paeoniae Radix Rubra),Danshen(Salviae Miltiorrhizae Radix et Rhizoma),Zhishi(Aurantii Fructus Immaturus),and Baizhu(Atractylodis Macrocephalae Rhizoma),were predicted for the treatment of CAD.Conclusion This study utilized bioinformatics and machine learning techniques to screen biomarkers for early diagnosis of CAD and targeted Chinese materia medica,providing a reference for the early prevention and treatment of CAD with TCM.