Objective: The aim of this study was to identify potentially important Rheumatoid arthritis (RA) targets related to immune cells based on bioinformatics analysis, and to identify small molecules of traditional Chinese medicine (TCM) associated with these targets that have potential therapeutic effects on RA. Methods: Gene expression profile data related to RA were downloaded from the Gene Expression Omnibus (GSE55235, GSE55457, and GSE77298), and datasets were merged by the batch effect removal method. The RA key gene set was identified by protein-protein interaction network analysis and machine learning-based feature extraction. Furthermore, immune cell infiltration analysis was carried out on all DEGs to obtain key RA markers related to immune cells. Batch molecular docking of key RA markers was performed on our previously compiled dataset of small molecules in TCM using AutoDock Vina. Moreover, in vitro experiments were performed to examine the inhibitory effect of screened compounds on the synovial cells of an RA rat model. Results: The PPI network and feature extraction with machine learning classifiers identified eight common key RA genes: MYH11, CFP, LY96, IGJ, LPL, CD48, RAC2, and CSK. RAC2 was significantly correlated with the infiltration and expression of five immune cells, with significant differences in these immune cells in the normal and RA samples. Molecular docking and in vitro experiments also showed that sanguinarine, sesamin, and honokiol could effectively inhibit the proliferation of RA rat synovial cells, also could all effectively inhibit the secretion of TNF-\alpha and IL-1\beta in synovial cells, and had a certain inhibitory effect on expression of the target protein RAC2. Conclusions: The core gene set of RA was screened from a new perspective, revealing biomarkers related to immune cell infiltration. Using molecular docking, we screened out TCM small molecules for the treatment of RA, providing methods and technical support for the treatment of RA with TCM.