Machine learning-based analysis of factors influencing surgical duration in type A aortic dissection

Figures

• 

  Figure 1. Data collection and preprocessing.

• 

  Figure 2. Distribution of surgical duration by gender.

• 

  Figure 3. Comparison chart of the real and predicted effects of all models on the test set. (a), Linear Regression. (b), Elastic Net. (c), Decision Tree Regressor. (d), Random Forest Regressor. (e), ExtraTrees Regressor. (f), Gradient Boosting Regressor. (g), XGBoost Regressor. (h), CatBoost Regressor. (i), AdaBoost Regressor. (j), Support Vector Regressor. (k), K-Nearest Neighbors Regressor.

• 

  Figure 4. Residual Distribution of all models on the test set. (a), Linear Regression. (b), Elastic Net. (c), Decision Tree Regressor. (d), Random Forest Regressor. (e), ExtraTrees Regressor. (f), Gradient Boosting Regressor. (g), XGBoost Regressor. (h), CatBoost Regressor. (i), AdaBoost Regressor. (j), Support Vector Regressor. (k), K-Nearest Neighbors Regressor.

• 

  Figure 5. Prediction Trend of all models on the test set. (a), Linear Regression. (b), Elastic Net. (c), Decision Tree Regressor. (d), Random Forest Regressor. (e), ExtraTrees Regressor. (f), Gradient Boosting Regressor. (g), XGBoost Regressor. (h), CatBoost Regressor. (i), AdaBoost Regressor. (j), Support Vector Regressor. (k), K-Nearest Neighbors Regressor.

• 

  Figure 6. Top 20 feature importance. (a), SHAP Beeswarm plot of the training set. (b), SHAP Bar plot of the training set. (c), SHAP Beeswarm plot of the test set. (d), SHAP Bar plot of the test set. (e), SHAP plot for a case study from the test set. (f), SHAP value for Duration of extracorporeal circulation. (g), SHAP value for Duration of aortic occlusion. (h), SHAP value for Intraoperative blood transfusion. (i), SHAP value for Treatment method for the aortic root. (j), SHAP value for Treatment method for the aortic arch. Each point represents a patient sample, and the colors from blue to red indicate the feature values from low to high.