In this study, we developed an innovative low-input strategy termed Iseq-Cit (internal standard-assisted enrichment-free approach for high-throughput quantitative analysis of citrullinome) to globally analyze the protein citrullinome in individuals at risk for RA who are asymptomatic and clinically healthy, yet test positive for ACPAs, as well as in RA patients within a longitudinal cohort study. Additionally, the citrullinome of 8 synovium samples from RA and osteoarthritis (OA) patients was profiled. Through bioinformatics analysis, we revealed the relationship between the citrullinomic landscape and RA severity. We also developed precise models based on clinical parameters and citrullination modifications to assess drug responsiveness in this longitudinal cohort. Furthermore, we created a bidirectional recurrent neural network for predicting antigenicity of citrullinated peptides, which were validated using enzyme-linked immunosorbent assay (ELISA) and T cell stimulation experiments.