Interpretable Machine Learning for Early Prediction of Prognosis in Sepsis: A Discovery and Validation Study

We have demonstrated the potential of machine-learning approaches for predicting outcome early in patients with sepsis. The SHAP method could improve the interpretability of machine-learning models and help clinicians better understand the reasoning behind the outcome.