Machine learning-based models for prediction of in-hospital mortality in patients with dengue shock syndrome

Figure 1 The study flow-chart. LASSO: Least Absolute Shrinkage and Selection Operator; DSS: Dengue shock syndrome; AUC: Area under the curve; AIC: Akaike information criterion; SMOTE: Synthetic minority oversampling technique; KNN: K-nearest neighbors.

Figure 2 Features of importance from the Random Forest model.

Figure 3 The SHAP analysis shows the impact of predefined variables on the model’s prediction of mortality in patients with dengue shock syndrome.