Fig. 10
Muc2 prediction model based on machine learning random forest method. Correlation between intestinal mucosal microbiota and targeted amino acid metabolism: A RDA analysis and B Procrustes Analysis. C A random forest classifier on the intestinal mucosal microbiota and serum amino acid quantification training sets to realize the cold and humid environmental stressed mice classification. (1) Decision tree learning algorithm. The markers were selected as key biomarkers by random forest method based on mean decrease Gini (2) and mean decrease accuracy (3). The red line illustrates the number of key markers in the discovery set. D Establishment of the prediction model. (1)A mathematical model based on 9 key markers was used to predict the sampling group. (2) The probability plot indicates the prediction probability. The red color indicates the predicted group and the grey color indicates the non-predicted group. (3) Taxa from the random forest model to establish the mathematical model could utilize the relative abundance of intestinal mucosal microbiota and serum metabonomic amino acid quantification to predict intestinal tissue Muc2 content with R2 of 0.75 and P of 0.012. CW-C, normal control group; CW-M, cold and humid environmental stress treatment group. CW-Cm, intestinal mucosa samples of normal control group; CW-Mm, intestinal mucosa samples of cold and humid environmental stress treatment group