Research Paper Volume 16, Issue 13 pp 10972—10984
Unveiling diabetic nephropathy: a novel diagnostic model through single-cell sequencing and co-expression analysis
Figure 5. Construction and validation of the diagnostic model for DN. (A) Identification of the four most significant genes (VEGFA, MAGI2, THSD7A, and ANKRD28) through feature screening. (B, C) Comparison of machine learning models, with the KNN method showing superior accuracy and AUC in distinguishing DN from control samples. (D–F) Performance evaluation of the diagnostic model in training and validation cohorts (GSE30122, GSE30528, and GSE30529) with AUC >0.9. (G, H) Nomogram construction for clinical application, enabling accurate prediction of DN risk based on gene expression.