Objective: Keloids are benign fibrotic conditions with increased proliferation, migration, and invasion, resembling tumor metastasis. Epithelial-mesenchymal transition (EMT) regulates tumor metastasis; thus, identifying EMT-related biomarkers may clarify keloid pathogenesis. Methods: Transcriptomic datasets (GSE83286, GSE212954, GSE90051) were analyzed using WGCNA, differential expression, and machine learning (LASSO, SVM-RFE, RF, XGBoost) to identify key EMT-related genes. Immune infiltration and single-cell analyses were performed, followed by experimental validation in human keloid tissues. Results: Twenty-nine EMT-related genes were identified. FN1 and RUNX2 emerged as key genes, with high diagnostic accuracy. Immune infiltration revealed reduced plasma cells, B cells, and dendritic cells, alongside increased resting mast cells and M2 macrophages in keloids. Single-cell and experimental validation confirmed FN1 and RUNX2 upregulation in keloid tissues. Conclusion: FN1 and RUNX2 are key EMT-related genes in keloids, offering potential diagnostic markers and therapeutic targets.