Purpose/Significance To integrate bioinformatics and machine learning approaches to identify immune-related key genes in Parkinson’s disease (PD)，and to characterize immune cell infiltration patterns and their potential roles in PD pathogenesis.Method/Process Based on the expression data of PD patients from the GEO database，candidate genes are screened through differential expression analysis，weighted gene co-expression network analysis (WGCNA) and immune gene sets from the ImmPort database. Four machine learning algorithms with cross-validation are applied to screen key genes and construct a logistic regression model. The model performance is evaluated using internal and external validation datasets. Single-sample gene set enrichment analysis is performed to characterize immune cell infiltration profiles.Result/Conclusion Two key genes(FGF13 and IL17RB) are identified，which may serve as potential diagnostic biomarkers for PD，providing new insights for clarifying the immune-related pathogenesis of PD.