Construction of a Diagnostic Model and Identification of Effect Genes for Diabetic kidney Disease with Concurrent Vascular Calcification Based on Bioinformatics and Multiple Machine Learning Approaches

Lili Huang¹Wenjing Wu²Shenhui Lv²Danfang Deng²Xiaoqin Wang^2*

• ¹Hunan University of Chinese Medicine, Changsha, China
• ²Hubei Provincial Hospital of Traditional Chinese Medicine, Wuhan, Hubei Province, China

Objective: This study aims to construct a diagnostic model for diabetic kidney disease (DKD) with concurrent vascular calcification (VC) using bioinformatics combined with machine learning approaches and to explore the potential underlying mechanisms. Methods: RNA sequencing (Bulk-seq) data of DKD and VC from various species were obtained from the Gene Expression Omnibus (GEO) database, and relevant datasets were integrated. Differential analysis of the DKD and VC datasets was performed using the limma package and weighted gene co-expression network analysis (WGCNA) in R (Ver. 4.3.3). Common differentially expressed genes (DEGs) and module genes were identified. Multiple machine learning algorithms were applied to select the optimal diagnostic model and identify hub genes, including LASSO regression, Random Forest, Gaussian Mixture Model (GMM), and Support Vector Machine-Reference (SVM-REF). Diagnostic performance was evaluated using the receiver operating characteristic (ROC) and precision-recall (PR) curves. Gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), gene set enrichment analysis (GSEA), and Cibersort immune infiltration analysis were conducted to explore the potential shared pathological mechanisms between DKD and VC.  Results: A total of five coDEGs (JUN, KCND3, HIP1, PTGDS, SLC22A17) were identified in our study. Among these three genes, JUN, PTGDS, and SLC22A17 demonstrated the best performance (validation group AUC: 1, test group AUC: 0.897) in the diagnostic model constructed by the SVM-REF machine learning method. Functional enrichment analysis of hub genes mainly involved biological processes such as inflammation, osteoblastic differentiation, apoptosis, and ferroptosis. Immune infiltration analysis revealed that in DKD patients, the expression levels of Memory B Cells, CD8+ T cells, M1 macrophages, M2 macrophages, resting dendritic cells, and resting mast cells were increased. In contrast, the expression of follicular helper T cells, activated mast cells, and neutrophils decreased relatively. Conclusion： This study suggests that JUN, PTGDS, and SLC22A17 may be potential biomarkers for DKD with VC, involving immune, metabolic, and inflammatory processes. These findings provide new targets for early diagnosis of DKD with VC and offer a novel perspective for applying bioinformatics combined with machine learning in discovering diagnostic biomarkers for diseases.