Oxidative stress-related genes as diagnostic markers for endometriosis and their associated immunoassays

Abstract

Purpose Endometriosis (EMT) affects millions of women worldwide and is closely associated with the body's response to oxidative stress. The role of oxidative stress markers in the diagnosis and treatment of endometriosis is a potentially fruitful avenue of research. Methods In this study, we employed a machine learning approach to model and screen key biomarkers, integrating five independent datasets from the Omnibus (GEO) database to construct a comprehensive training set. The identification of key genes was achieved through a process of cross-referencing with the aim of locating those that were differentially expressed and known to be involved in oxidative stress. Nine machine learning algorithms were employed for model selection, followed by the evaluation of immune infiltration and immune correlation through single sample gene set enrichment analysis (ssGSEA) and the CIBERSORT algorithm. Results After comparing the performance of different machine learning algorithms, the gradient boosting algorithm (GBM) was selected as the best model. Eventually it screened five featured genes (FOS, CFH, AOX1, FMO1, FCGR2B). The expression patterns of these genes showed diagnostic and predictive potential in the constructed nomograms and external validation. In addition, the association of these genes with pregnancy status and disease severity was explored. The results of immune infiltration analysis showed significant correlation between these key genes and the immune system. Conclusions This study identifies genes at the intersection of endometriosis and oxidative stress, thereby providing reliable molecular markers for clinical application. This offers a new avenue for subsequent diagnosis and treatment of endometriosis.