Fig. 1

Workflow of the construction of the XGBoost-based classification model. After the IBD dataset was collected, the XGBoost algorithm and UMAP were employed to select important features (32-gene signature). Then, ten-fold cross-validation tests were set to compare the performance of the models between two feature sets. Finally, The XGBoost method was used to compare the performance of the XGBoost-based classification model on unused data and predict the probability of IBD for each case