This study used integrated bioinformatics and machine learning to identify genes that may distinguish rheumatoid arthritis (RA) from osteoarthritis (OA) using publicly available synovial tissue transcriptomic data. By applying LASSO and support vector machine–recursive feature elimination and selecting overlapping results, the researchers identified three candidate genes—EPYC, MAGED1, and LAP3—that demonstrated good discriminatory performance (AUC > 0.85) across two datasets. These genes were associated with immune-related pathways and immune cell infiltration patterns, and their expression changes were partially validated in a tumor necrosis factor alpha–stimulated fibroblast-like synoviocyte model. However, the findings are based on small, cross-sectional datasets with limited experimental validation, and further clinical and mechanistic studies are required before diagnostic application.