Fig. 7

Comparison of predictors. a Omes by model, with c = CANTARE, p = penalized regression, r = random forest. The number of predictors included is shown in parentheses after the model name. Each circle represents the inclusion of age, fecal calprotectin, or any of 3 omes (e = enzyme, mb = microbe, met = metabolite) in each of 15 models. For random forests, the ten most important variables are considered. Circle size represents the percent of predictors/important variables drawn from that ome. b Included predictors/important variables by model. All variables that are included as predictors or identified as important are listed. Circles indicate which predictors are included in which models. Predictors flagged with an asterisk (*) are metabolic isomers. After age and calprotectin, variables are listed alphabetically, within ome