Using Long Short-Term Memory (LSTM) Neural Networks to Predict Emergency Department Wait Time

Area of Science:

• Health Informatics
• Artificial Intelligence in Healthcare
• Operations Research

Background:

• Emergency Department (ED) overcrowding is a significant global healthcare challenge.
• Accurate prediction of patient wait times is essential for improving ED efficiency and patient satisfaction.
• Machine learning models offer potential solutions for predicting ED wait times.

Purpose of the Study:

• To develop and evaluate a Long Short-Term Memory (LSTM) recurrent neural network model for predicting emergency department wait times.
• To compare the predictive performance of the LSTM model against a Linear Regression (LR) model.

Main Methods:

• Utilized a randomly generated patient timestamp dataset simulating a typical hospital journey.
• Implemented a Long Short-Term Memory (LSTM) recurrent neural network for time-series prediction.
• Employed a Linear Regression (LR) model as a benchmark for comparison.

Main Results:

• The LSTM model demonstrated a statistically significant reduction in average mean absolute error by 9.7% (3 minutes) compared to the LR model (p < 0.01).
• The LSTM model achieved superior predictive accuracy for emergency department wait times.
• Both LSTM and LR models showed practical utility in predicting ED wait times.

Conclusions:

• Long Short-Term Memory (LSTM) networks provide a more accurate approach to predicting emergency department wait times.
• Implementing advanced machine learning models like LSTM can enhance patient experience and operational efficiency in emergency departments.
• Further research can explore integrating these models into real-time ED management systems.