Skip to main content
Fig. 5 | Microbiome

Fig. 5

From: Characterization of TMAO productivity from carnitine challenge facilitates personalized nutrition and microbiome signatures discovery

Fig. 5

Selected TMAO-productivity-related microbiome features were used to build a TMAO producer prediction model. a Random forest classification modelling was built using the 39 selected OTUs, yielding a higher prediction accuracy (AUROC = 0.81) than that obtained using all 1637 OTUs (AUROC = 0.79). The prediction model was then successfully validated with an external cohort of patients with CVD (AUROC = 0.80). The 95% confidence interval of the AUCs is shown in parentheses. The optimal cutoff points were marked on the ROCs. b Variable importance plot of the 39 selected features ranked by mean decrease accuracy. The E. timonensis was ranked fifth among the 39 annotated taxa

Back to article page