Fig 1.
Distribution of data.
Fig 2.
Workflow of the model.
Fig 3.
ROC-AUC curve of models implemented.
Fig 4.
Confusion Matrix for XGBoost Classification Model.
Fig 5.
Confusion Matrix for Random Forest Classification Model.
Fig 6.
Classification metrics for XGBoost and RF: The table below depicts the values of metrics for XGBoost and Random Forest model.
Fig 7.
Comparison of both models: It can be understood that the XGBoost outperforms Random Forest in terms of classification metrics.
So, we have chosen XGBoost as our final model to be used in the interface. The interface helps the user to predict the possibility of postoperative delirium.