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

Abstract

Cow's milk protein allergy (CMA) is one of the most common food allergies in children worldwide. However, it is still not well understood why certain children outgrow their CMA and others do not. While there is increasing evidence for a link of CMA with the gut microbiome, it is still unclear how the gut microbiome and metabolome interact with the immune system. Integrating data from different omics platforms and clinical data can help to unravel these interactions. In this study, we integrate clinical, microbial, (meta)proteomics, immune and metabolomics data into machine learning (ML) classification, using multi-view learning by late integration. The aim is to group infants into those that outgrew their CMA and those that did not. The results show that integration of microbiome data with clinical, immune, (meta)proteomics and metabolomics data could considerably improve classification of infants on outgrowth of CMA, compared to only considering one type of data. Moreover, pathways previously linked to development of CMA could also be related to outgrowth of this allergy.