Screening and validation of ZFYVE27 as a potential diagnostic biomarker for osteoporosis via integrative bioinformatics and machine learning approaches

PurposeOsteoporosis (OP) is a systemic metabolic skeletal disorder characterized by diminished bone mineral density, deteriorated bone microarchitecture, and a consequently heightened susceptibility to fragility fractures. Bioinformatics approaches serve as a crucial bridge between genomic investigations and clinical translation, and have been extensively utilized in OP research.

Nevertheless, the precise identification of core pathogenic genes and the subsequent development of robust and accurate diagnostic biomarkers remain urgent clinical imperatives.MethodsInitially, we identified differentially expressed genes (DEGs) by comparing transcriptomic profiles between healthy controls and osteoporotic patients, followed by functional enrichment analyses of associated biological processes and signaling pathways. Weighted gene co-expression network analysis was subsequently applied to isolate disease-specific module genes.

By intersecting these module genes with the DEGs, OP-related DEGs were precisely delineated. The LASSO regression algorithm was utilized to filter seven hub candidate genes.

Subsequently, Support Vector Machine and Random Forest machine learning algorithms were employed to further optimize and cross-validate these potential diagnostic biomarkers. The intersection of these multi-algorithmic outputs ultimately designated the core biomarker.