Hot Topics in Neonatology
December 6-10, 2020
View the Full Poster
Introduction
Numerous studies have sought to find associations among microbial impact factors, clinical practices, and the state of the preterm infant gut microbiome. Current longitudinal studies are limited in their ability to integrate the microbiome analysis with clinical data. In this study, we hypothesized that integrated longitudinal visualizations of the microbial health risk factors and microbiome analysis could create novel insights and observations regarding the impact of clinical decisions on the infant gut health.
Methods
We developed a dashboard to provide a “visual narrative” for individual patients. We leveraged a large retrospective dataset of chronologically sampled infants (n=267) born under 34 weeks, sequenced using shotgun metagenomic sequencing (n=2996 samples, Illumina NextSeq 2×150). The dashboard presents perinatal data and birth facts integrated with feeding rates, diets, medications, diagnoses, z-score for weight, and selected microbiome-related measures. This representation includes relative abundance, Preterm Gut Community Types (PGCT)1, and microbiota for age z-score (MAZ)2. Machine learning predictions for outcomes were incorporated as well. Clinical data algorithms were trained on a superset of infants (n=417). Predicted outcomes include growth failure defined as a birth-to-discharge z-score decline of <1.2, and non-typical development, defined as growth failure, NEC, and/or sepsis.
Results
All data was integrated and aligned by day of life allowing researchers to visualize microbiome changes in relation to clinical decision making (Fig. 1).
Conclusion
To develop intervention strategies that promote a healthier infant gut, an evidence-based tool is needed to establish standard measurements of the infant gut, along with longitudinal views integrating clinical data.
- Tandon, D. Genetti, J. Levesque, A. Alicea, A. Nayeem, K. Lee, D. Gallagher, T. Warren, Astarte Medical
