Abstract
Microbiome-based disease prediction shows promise as an early, noninvasive indicator of various health conditions linked to gut dysbiosis, aided by declining sequencing costs. Current microbiome health indices largely rely on species richness and taxonomic classification. A renewed focus on a metabolism-centric, ecological approach has revealed limitations of these traditional methods. This study presents a new metagenomic health index aimed at distinguishing healthy from unhealthy microbiomes, specifically in inflammatory bowel disease (IBD). Our approach shifts from Linnean phylogenetic classification to a functional analysis of metabolic potential and ecological interactions among species. We compare our index's performance against established indices, including the Gut Microbiome Health Index (GMHI) and high-dimensional principal component analysis (hiPCA), using data from IBD cohorts and 27 additional clinical datasets. Our index outperforms GMHI and hiPCA in identifying health states, particularly in a longitudinal COVID-19 cohort, and demonstrates robustness to sequencing depth. We advocate for a functional approach to better assess microbiome health and suggest future index improvements.