A multi-omics machine learning classifier for outgrowth of cow's milk allergy in children

Area of Science:

• Pediatric Allergy and Immunology
• Gut Microbiome Research
• Computational Biology

Background:

• Cow's milk protein allergy (CMA) is a prevalent childhood condition with unpredictable resolution.
• The role of the gut microbiome in CMA development and resolution is increasingly recognized but not fully understood.
• Interactions between the gut microbiome, metabolome, and immune system in CMA are complex and require integrated analysis.

Purpose of the Study:

• To investigate the factors influencing the outgrowth of cow's milk protein allergy in infants.
• To determine if integrating multi-omics data with clinical information can improve classification of CMA outgrowth.
• To identify key microbial and metabolic pathways associated with CMA resolution.

Main Methods:

• Integration of diverse datasets including clinical, microbial, (meta)proteomics, immune, and metabolomics data.
• Application of machine learning, specifically multi-view learning by late integration, for classification.
• Comparative analysis of classification performance using single-view versus integrated multi-view data.

Main Results:

• The integration of gut microbiome data with clinical, immune, (meta)proteomics, and metabolomics data significantly enhanced the classification accuracy of infants who outgrew CMA.
• Multi-view learning demonstrated superior performance compared to models relying on single data types.
• Identified specific biological pathways linked to both CMA development and its subsequent outgrowth.

Conclusions:

• Integrated multi-omics data analysis is a powerful approach for understanding complex allergic diseases like CMA.
• Gut microbiome composition and function play a crucial role in determining CMA resolution in infants.
• This study provides a foundation for developing predictive biomarkers for CMA outgrowth.