"VALUTAZIONE DELL'EUBIOSI/DISBIOSI DEL MICROBIOMA INTESTINALE BASATA SU DATI DI DNA METABARCODING MEDIANTE APPROCCI DI APPRENDIMENTO AUTOMATICO"

The human body harbors trillions of microbes, widely known as the human microbiome. Over the past 20 years, the advent of next-generation sequencing (NGS) technologies and the progressive decrease of sequencing and analysis costs has laid the basis to understand the tight relationship among the human microbiome and host health. To date, several findings in scientific literature highlight how the human microbiota, especially the gut one, orchestrates several crucial biological processes underpinning host health. Although it is currently clear that environmental factors and host genetics strongly influence the composition of the gut microbiome, it is also clear that a high diversity of microbial communities (eubiosis) is closely associated with overall human well-being. On the other hand, conditions of reduced diversity (dysbiosis) seem to be closely associated with a pro-inflammatory microenvironment. Although microbiota research has rapidly grown, what defines a 'healthy microbiome' remains to be clarified. Given the high dimensionality, heterogeneity and poor availability of metadata associated with microbiome data, machine learning approaches may provide a powerful tool to untangle the complexity of microbiome-human host interactions, circumventing the limitations of traditional statistical methods. My PhD project aims to develop an eubiosis/dysbiosis model based on machine learning approaches using DNA metabarcoding data.