Neural Network and Random Forest Algorithms for Estimation of the Waiting Times Based on the DES in ED

Abstract

This study aimed to predict patient waiting times for an ED (emergency department) unit by integrating DES (discrete event simulation) model and ML (machine learning) algorithms. The health resources in the DES model were kept constant. However, the results were obtained by including the statistical distributions of the processes in the DES. Length of stay (LOS), resource efficiency rates, patient genders, walking distance, time of processes, and age were considered input factors that affect patient waiting times. Prediction data were calculated using Neural Network (NN) and Random Forest (RF) models from ML algorithms. Testing and training phases of ML algorithms are set to 20% and 80%. The RF model performed best with low RMSE, MSE, MAE, and high R2 values. This model's RMSE, MSE, MAE, and R 2 values were calculated as 2.81, 1.67, 0.88, and 0.996, respectively. As a result, this study suggested integrating DES and ML models to overcome many factors, such as satisfaction, cost, and quality, with the intense human factor in service sectors with dynamic structures.