OBJECTIVES: Little is known about maturation of the airway microbiota during early childhood and the consequences of early-life antibiotic exposure.
METHODS: In a population-based birth cohort of 902 healthy Finnish children, we applied deep neural network models to investigate the relationship between the nasal microbiota (measured by 16S rRNA gene sequencing at up to three time points) and child age during the first 24 months. We also performed stratified analyses according to antibiotic exposure during the age period 0-2 months.
RESULTS: The dense deep neural network analysis successfully modelled the relationship between 232 bacterial genera and child age with a mean absolute error of 4.3 (95%CI 4.0-4.7) months. Similarly, the recurrent neural network analysis also successfully modelled the relationship between 215 genera and child age with a mean absolute error of 0.45 (95%CI 0.42-0.47) months. Among the genera, Staphylococcus spp. and members of the Corynebacteriaceae decreased with age, while Dolosigranulum and Moraxella increased with age in the first 2 years of life (all false discovery rate (FDR) = 0.001). In children without early-life antibiotic exposure, Dolosigranulum increased with age (FDR = 0.001). By contrast, in those with early-life antibiotic exposure, Haemophilus increased with age (FDR = 0.002).
CONCLUSIONS: In this prospective birth cohort of healthy children, we demonstrated the development of the nasal microbiota, with shifts in specific genera constituting maturation, in the first 2 years of life. Antibiotic exposures during early infancy were related to different age-discriminatory bacteria.
Keywords: Airway Antibiotics Children Deep neural network model Infant Nasal microbiota STEPS study