Abstract
OBJECTIVES: To establish a ferroptosis-related colorectal cancer (CRC) diagnostic model by integrating machine learning and gene expression analysis. Tumorigenesis is strongly linked to ferroptosis, an iron-mediated kind of cell death. The dismal prognosis of CRC, a severely malignant gastrointestinal cancer, accentuates the demand for effective diagnostic biomarkers. METHODS: The study was performed between January 2024 and 2025. The GEO database provided 2 openly searchable gene expression profiles (GSE9348 and GSE21510) from CRC as well as non-tumor tissues. Genes that were expressed differently in tumor and healthy tissues were found using the GSE9348 dataset. Distinct genetic biological functions were identified through functional enrichment analysis. SVM-RFE and LASSO regression models helped identify potential genetic markers associated with CRC. RESULTS: GSE9348 dataset analysis helped identify 27 differentially expressed ferroptosis-related genes. KEGG analysis suggested that these genes are primarily related to inflammatory responses, NF-κB signaling, and regulation of the interleukin family. Based on this model, CHMP2A, CYCS, HMGB1, IL18, IL1B, and GZMA were selected as potential diagnostic markers, and a novel diagnostic model was constructed. Its predictive value was examined using receiver operating characteristic analysis. We validated the expression changes of model genes using PCR assays along with a validation set (comprising TCGA and GSE21510 datasets). CONCLUSION: These outcomes provide an efficient ferroptosis-related gene-based diagnostic model for CRC. Nevertheless, before its use in real-time settings, more clinical studies are required to confirm its diagnostic value.