Skip to main content
Fig. 5 | Gut Pathogens

Fig. 5

From: Using whole-genome sequencing (WGS) to plot colorectal cancer-related gut microbiota in a population with varied geography

Fig. 5

CRC risk prediction model and importance of variables. Figure A Overall CRC risk prediction model based on the genus and species levels in all regions. Figure A-a1: ROC curve at the genus and species levels. Figure A-a2: Top 20 characteristic model interpretation diagrams. The left side represents the relative weights of the corresponding features in the 15 cross-validation submodels in the training model, the middle is the normalized abundance values of each species among the grouped samples, and the right side is the boxplot of the ratio of the 20 features with nonzero coefficients among the 15 submodels. Figure B Cross-validation of disease risk prediction models between species datasets in different regions. Figure C Venn diagram of the top 20 characteristic bacteria between datasets at the species level. Figure D Top 20 characteristic model interpretation diagrams of the species-level CRC risk prediction model in each region in the map

Back to article page