Prediction of Waiting Times in A&E

Area of Science:

• Emergency Medicine
• Artificial Intelligence
• Data Science

Background:

• Accurate prediction of waiting times in emergency departments (A&E) is crucial for effective patient flow management.
• Traditional methods like rolling averages fail to capture the complex contextual factors influencing A&E wait times.
• The pre-pandemic period (2017-2019) provides a valuable dataset for developing robust predictive models unaffected by recent global health events.

Purpose of the Study:

• To develop and evaluate an AI-enabled method for predicting patient waiting times in A&E departments.
• To compare the performance of machine learning algorithms against conventional prediction techniques.
• To establish a more dynamic and accurate system for forecasting patient discharge times prior to arrival.

Main Methods:

• Utilized retrospective A&E patient data from 2017-2019.
• Trained and tested Random Forest and XGBoost regression models to predict time to discharge.
• Applied models to a dataset of 68,321 patient observations using a comprehensive feature set.

Main Results:

• The XGBoost model achieved a Root Mean Square Error (RMSE) of 82.66 and a Mean Absolute Error (MAE) of 64.31.
• The Random Forest model demonstrated strong performance with RMSE=85.31 and MAE=66.71.
• Both AI models significantly improved predictive accuracy compared to standard methods.

Conclusions:

• AI-enabled regression models, specifically XGBoost and Random Forest, offer a more dynamic and accurate approach to predicting A&E waiting times.
• These advanced methods provide valuable insights for optimizing patient flow and resource allocation in emergency departments.
• The findings suggest a shift towards data-driven, predictive analytics for enhanced operational efficiency in healthcare settings.