Deciphering immune-inflammatory dysregulation in the endometriotic microenvironment: insights from single-cell omics and artificial intelligence.

Endometriosis is a prevalent chronic inflammatory gynecological disorder affecting approximately 10% of reproductive-age women worldwide, characterized by endometrial-like tissue outside the uterine cavity. Ectopic lesion growth tracks closely with immune-inflammatory dysregulation-altered macrophage polarization, impaired natural killer (NK) cytotoxicity, skewed T cell subsets, B cell-related autoimmunity, tolerogenic dendritic cells, mast cell-associated neuroinflammation, and abnormal cytokine networks. Even after many years of study, several regulatory mechanisms in the endometriotic microenvironment remain only partly defined. Single-cell omics-especially single-cell RNA sequencing (scRNA-seq), spatial transcriptomics, mass cytometry (CyTOF), and multi-omics integration-maps immune composition and cell-cell communication at a level bulk assay typically miss, including rare states and niche structure. Artificial intelligence (AI) and machine learning (ML), including single-cell foundation models, deep learning for drug repurposing, immune deconvolution, and large language models, are now common choices for integrating large datasets, deriving immune signatures, ranking candidate targets, and supporting translation. This review summarizes recent work at that interface: immune heterogeneity and dysfunction across macrophage, NK, T, B, dendritic cell, and mast cell compartments; AI-assisted biomarker studies, repurposing, and network pharmacology, including natural products and traditional Chinese medicine; and practical limits that still affect clinical application.