Development and external validation of an interpretable machine-learning model for HFpEF comorbidity risk in COPD patients

Figures

• 

  Fig. 1. Flowchart of study procedure. SHAP: SHapley Additive exPlanatio

• 

  Fig. 2. Interaction between variables in a correlation matrix

• 

  Fig. 3. Testing and evaluation of eXtreme Gradient Boosting (XGBoost) model based on internal validation data. A Area under the receiver operating characteristic curve comparison between models; B The confusion matrix of XGBoost model; C Comparison of decision curve analysis curves between models; D Comparison of calibration curves between models. AUC: Area under the receiver operating characteristic curve; ROC: Receiver operating characteristic; SVM: Support Vector Machine (SVM); GBM: Gradient boosting machines; KNN: K-nearest neighbors; LightGBM: Light gradient boosting machine; XGBoost: eXtreme Gradient Boosting (XGBoost)

• 

  Fig. 4. Testing and evaluation of eXtreme Gradient Boosting (XGBoost) model based on external validation data. A Area under the receiver operating characteristic curve comparison between models; B The confusion matrix of XGBoost model; C Comparison of decision curve analysis curves between models; D Comparison of calibration curves between models. AUC: Area under the receiver operating characteristic curve; ROC: Receiver operating characteristic; SVM: Support Vector Machine (SVM); GBM: Gradient Boosting Machine (GBM); KNN: K-nearest neighbors; LightGBM: Light gradient boosting machine; XGBoost: eXtreme Gradient Boosting (XGBoost)

• 

  Fig. 5. Interpretation of the eXtreme Gradient Boosting (XGBoost) model using SHapley Additive exPlanations. A Mean importance ranking of features displayed by SHapley Additive exPlanations (SHAP); B Bee colony diagram of characterization attributes in SHAP; C Waterfall plot of individual SHAP analysis for patient #112 with COPD and HFpEF; D Partial dependence plot for the interaction between CAT score and NT-proBNP; E Force plot of individual SHAP analysis for patient #216 with COPD and HFpEF; F The SHAP force plot for patient #300 with COPD and HFpEF shows exclusively positive feature contributions

• 

  Fig. 6. Demonstration of a Shiny-deployed XGBoost prediction model based on a patient with COPD and HFpEF