CFE 2019: Start Registration
View Submission - CMStatistics
Title: Probabilistic forecasting of patient waiting times in the emergency department Authors:  Siddharth Arora - University of Oxford (United Kingdom) [presenting]
James Taylor - University of Oxford (United Kingdom)
Abstract: Accurate estimates of patient waiting times in the emergency department (ED) have been associated with increased patient satisfaction and improved outcomes. Moreover, waiting time estimates can assist hospitals to streamline patient-flow based on informed staff and resource allocation. Individual patient waiting times in ED are inherently uncertain. We thus generate and evaluate probabilistic forecasts, based on the following categories of predictor variables: (1) workload, (2) staffing, (3) calendar variables, (4) demographics, and (5) severity of the patient condition. Using around 350,000 anonymized patient-level ED records collected over a period of five years for one of the major hospital sites in the UK, we develop a methodology to: (1) predict patient waiting times for both major and minor triage categories, (2) identify the variables that have the highest impact on modelling accuracy, and (3) accommodate the dynamic nature of patient-flow in ED via re-estimation of predictor variables. Out-of-sample point and probability distribution forecasts are evaluated using the mean absolute error (MAE) and continuous ranked probability score (CRPS), respectively.