We developed an ML model based on XGBoost. The ML model could provide effective and efficient differential diagnoses of ITB and CD with diagnostic bases. The ML model performs well in real-world clinical practice, and the agreement between the ML model and MDT is strong.