Machine learning for identifying core biomarkers of immune dysregulation in patients with recurrent implantation failure

Objective: The etiology and pathogenesis of recurrent implantation failure (RIF) remain unclear. This study aimed to delineate the immune landscape and identify the pivotal biomarkers of immune dysregulation in RIF. Methods: Comprehensive bioinformatics analyses, including differential expression, functional enrichment, gene set enrichment, weighted gene co-expression network analysis, protein–protein interaction network, and machine learning analyses, were conducted to explore the immunological underpinnings of RIF. Results: Our findings demonstrated a significant downregulation in the expression of several immune-related gene sets in the RIF cohort, including those related to antigen processing and presentation, antimicrobial responses, chemokines and their receptors, cytokines, and natural killer cell cytotoxicity. These results provide evidence of a significant association between RIF and maternal–fetal immune dysregulation. Three signature genes ( ANK3 , PTGS2 , and SLC39A2 ) were identified using various machine-learning methods. The diagnostic value and immunological relevance of these genes were confirmed in both discovery and validation cohorts. Conclusion: This study investigated the immune landscape of RIF using extensive bioinformatic analyses. Modular genes associated with both immune activity and RIF dysregulation were identified, providing new perspectives for understanding the pathophysiology and treatment of RIF.