Abstract: Objective To identify key metabolic genes in tuberculosis (TB) patients and to explore their diagnostic value. 
Methods Datasets GSE83456 and GSE42834 were downloaded from the GEO database. Differential expression analysis, pathway enrichment analysis, and machine learning algorithms were employed to identify and analyze key metabolic genes in TB. 
Results A total of 1 170 differentially expressed genes (DEGs) were identified between TB patients and healthy controls, primarily enriched in immune-related pathways. After intersecting with metabolic gene sets, five key metabolic genes (PRDX6, MGLL, RENBP, WASF3, and IDO1) were screened using machine learning algorithms. A neural network model demonstrated high predictive accuracy for these genes. 
Conclusion Through comprehensive bioinformatics analysis, five key metabolic genes are identified, providing new insights for the early diagnosis and treatment of TB. 

Key words: tuberculosis, metabolic genes, diagnosis