Identification of ferroptosis-genes associated with pediatric inflammatory bowel disease bioinformatics and machine learning approaches.

Background: Pediatric inflammatory bowel disease (PIBD) is increasingly common, and early diagnosis remains challenging due to unclear etiology. Ferroptosis, an iron-dependent form of cell death, may be involved in intestinal inflammation, but its expression and role in PIBD are poorly understood. Objective: To identify ferroptosis-related genes as candidate biomarkers for early diagnosis of PIBD and validate their role in ferroptosis. Methods: RNA-seq data of PIBD from GEO datasets were analyzed using DESeq2, WGCNA, and functional enrichment analysis. Ferroptosis-related diagnostic genes were screened through LASSO, Random Forest, and mSVM-RFE algorithms, and validated in GSE57945 and GSE117993 datasets. Results: PML and CHAC1 were identified as potential biomarkers for early diagnosis of PIBD, showing high diagnostic performance (AUC > 0.7) in training, validation, and external datasets. Conclusion: PML and CHAC1 are promising biomarkers for early PIBD diagnosis. These findings, supported by both bioinformatic analyses and experimental validation, may improve diagnostic accuracy and provide insights into the immune microenvironment and therapeutic strategies.