A comparison of interpretable XGBoost and artificial neural network model for the prediction of severe acute pancreatitis

What's new?Acute pancreatitis (AP) is known as one of the gastrointestinal emergency conditions.A certain proportion of the AP patients develop into severe acute pancreatitis (SAP) with high mortality.Early identification of SAP and adoption of treatment measures have a positive impact on improving patient prognosis.This study developed and compared two types of machine learning models, including Extreme Gradient Boosting (XGBoost) and artificial neural network model (ANN), predicting of SAP.As far as we know, it's the first publication to provide an interpretable XGBoost model with local interpretable model-agnostic explanations (LIME) diagrams for predicting SAP development.Moreover, we identified important predictors of SAP.In view of high identification efficiency of XGBoost model for SAP, clinicians could recognize severe patients early and then correspondingly take active measures.