Fig. 7

The performance of classification models based on fecal microbiota CAGs and metabolite modules in discriminating non-return sows from normal sows. A Aera under the receiver operating characteristic curve (AUC) of three random forest models with biomarkers of fecal microbiota CAGs and metabolite modules. The diagnostic performance was assessed using receiver operating characteristic analysis. Colors represent different models. Cl, confidence intervals. B Ten folds in random forest cross-validation for feature selection C The classification model with the best diagnostic performance is based on the gut microbiota CAGs and metabolite modules. The length of the line represents the importance of biomarkers. The red dots represent biomarkers significantly enriched in non-return sows, while blue dots represent biomarkers significantly enriched in normal sows