Abstract / Summary
Abstract Childhood obesity (CO) is a complex chronic disease driven by environmental, behavioral, and genetic factors. Dysregulation of branched-chain amino acid metabolism (BCAAM) contributes to CO development, highlighting the need to identify reliable biomarkers for diagnosis and treatment. Biomarkers associated with BCAAM remain largely unexplored in pediatric populations. In this study, we used BCAAM-focused co‑expression network analysis with machine learning to identify biomarkers and construct a diagnostic nomogram for CO using transcriptomic data from blood and adipose tissue. Multidimensional analyses, including functional enrichment, immune infiltration, and drug prediction, were subsequently performed. Three putative biomarkers, PLEK, NIN, and COX1, were identified, and the resulting nomogram exhibited good predictive performance. These biomarkers showed strong transcriptional association with mitochondrial respiration pathways and immune cell infiltration signatures, suggesting mitochondria–immune crosstalk. We also predicted multiple candidate drugs that target these biomarkers, with quercetin among the top candidates. Validation of their expression by RT‑qPCR in clinical blood samples confirmed significant dysregulation of PLEK and COX1, which supports their potential as biomarker candidates. These results indicate that PLEK, NIN, and COX1 are candidate BCAAM-associated transcriptomic markers for CO. The nomogram shows promising diagnostic potential in retrospective datasets; however, its clinical utility and the underlying mechanistic framework require further study.