Integrating early-life gut microbiota with maternal–infant clinical data to classify mild head-circumference restriction at one month

Early head circumference is a vital proxy for neurodevelopment. We investigated whether the one-month gut microbiome adds predictive value to clinical factors for mild growth restriction. We analyzed anthropometry and stool 16S rRNA gene sequencing from 161 mother–infant dyads in a prospective cohort. Four machine-learning models classified mild restriction (Z-score < −1) using clinical and genus-level microbiome features, evaluated via repeated stratified cross-validation to assess incremental predictive value. Restricted infants (n = 14) exhibited lower microbial richness, reduced Bifidobacterium , and taxonomic shifts. Clinical-only LightGBM achieved the highest AUC overall (0.808), whereas microbiome-only models showed weaker discrimination (AUC 0.432–0.636). A small, non-significant AUC increase was observed for elastic-net logistic regression after microbiome features were added (0.761 → 0.802, 95% CI including zero). The other three learners showed lower combined-model AUCs, including a significant decrease for LightGBM; none of the four paired comparisons showed a statistically significant improvement. Mild head-circumference restriction at one month was associated with lower microbial richness and altered gut ecology. Although clinical anthropometry provided the strongest classification signal in this cohort, the concurrent richness association is biologically plausible and supports continued investigation of the early-life microbiome as a candidate contributor to growth restriction in larger, prospectively sampled cohorts.