AI-Based Exhaled VOC Panels as Gut Microbiota-Derived Biomarkers for IBD Screening

Inflammatory bowel diseases (IBD) present significant diagnostic and therapeutic challenges due to their heterogeneity. Here, we analyze exhaled volatile organic compounds (VOCs) from 244 individuals, uncovering distinct metabolic signatures associated with IBD. Using advanced thermal desorption gas chromatography-tandem mass spectrometry, we identified 64 endogenous metabolites, including elevated hydrocarbons, and depleted organic acids and derivatives, as well as lipids and lipid-like molecules. Integrating gut microbiota data from 1660 individuals, we link these metabolic shifts to microbial dysbiosis, characterized by reduced beneficial bacteria and heightened oxidative stress. Machine learning identified a 10-VOC biomarker panel, achieving superior diagnostic accuracy (AUC = 0.88) across IBD subtypes and activity states. A novel risk scoring system further improved diagnostic precision. Probiotic interventions in colitis models confirmed that restoring microbial balance, particularly via butyrate production, mitigates inflammation and repairs intestinal barriers. These findings position breathomics as a groundbreaking tool for IBD diagnosis, stratification, and microbiota-targeted therapies.