An Interpretable Machine Learning Model for Predicting in-Hospital Progression in Initially Mild...

Area of Science:

• Medical Imaging
• Machine Learning
• Gastroenterology

Background:

• Hypertriglyceridemia-induced acute pancreatitis (HTG-AP) can have unpredictable disease courses.
• Early identification of patients at risk for progression is crucial for timely intervention.

Purpose of the Study:

• To develop an interpretable machine learning model for early prediction of in-hospital progression in mild HTG-AP.
• To utilize baseline clinical and non-contrast computed tomography (CT) features for risk stratification.

Main Methods:

• Retrospective enrollment of 164 patients with mild HTG-AP.
• Collection of clinical, CT-based body composition, and pancreatic radiomics features.
• Development and validation of four XGBoost models, including a multimodal approach (clinical + body composition + radiomics).

Main Results:

• The multimodal model (clinical + body composition + radiomics) showed the best performance (AUC 0.830, accuracy 0.768).
• Triglycerides (TG) and visceral fat area-to-abdominal cavity area ratio (VFA/ACA) were key predictors.
• The model demonstrated good calibration and clinical utility via decision curve analysis.

Conclusions:

• An interpretable multimodal machine learning model can improve early risk stratification for HTG-AP.
• This approach may help identify patients needing closer monitoring.
• External validation is recommended to confirm generalizability.