Explainable artificial intelligence advances breath analysis for monitoring Crohn’s disease

Breath analysis has emerged as a promising tool for non-invasive disease monitoring, particularly for tracking ongoing disease, due to its reproducibility and accessibility. One of the disorders that can be investigated through breath analysis is inflammatory bowel disease, such as Crohn’s disease. While many studies address this disease epidemiologically, at present a comprehensive understanding of biomarkers that can differentiate the metabolic responses between male and female patients is still missing. In this study, we explore significant biomarkers in classifying male and female body response with Crohn’s disease using explainable artificial intelligence. By leveraging a public dataset, we apply an AI-based algorithm to analyze the volatile organic compounds present in exhaled breath, which impact gender classification. Our data-driven approach not only identifies key endogenous biomarkers, but also provides interpretability in understanding patterns useful to reveal a differentiated metabolic response and develop personalized clinical treatment. The results highlight the potential of AI in advancing breath analysis for personalized monitoring of Crohn’s disease and improving gender-specific medicine.