Skip to main content
. 2022 Aug 23;10:893059. doi: 10.3389/fped.2022.893059

Figure 1.

Figure 1

Machine learning revealed NEC microbiome signatures by post-menstrual age. (A) Pathway level data was stratified by species, oversampled to balance NEC classes, sub-setted by PMA, and trained on different model algorithms to explore model performance. Models sub-setted to include only samples from subjects with a PMA ≥29 weeks performed moderately better with respect to specificity and sensitivity than their counterparts. (B) Principal coordinate analysis (PCoA) of all samples based on weighted UniFrac distance and colored by Necrotizing Enterocolitis (NEC) or preterm infants without NEC (Preterm).