Data-driven queueing modelling: a simulation case study of emergency department crowding.

Area of Science:

• Healthcare Operations Research
• Emergency Medicine Management
• Computational Statistics

Background:

• Emergency department crowding is a complex issue impacting patient care and operational efficiency.
• Existing modeling methods for crowding present a dichotomy between predictive accuracy (machine learning) and evaluative capabilities (queueing/simulation).
• A significant gap exists in methodologies that can both predict crowding and evaluate the impact of operational changes.

Purpose of the Study:

• To implement and validate a novel data-driven queueing methodology for emergency department crowding.
• To bridge the gap between machine learning prediction and queueing/simulation evaluation methods.
• To demonstrate the methodology's applicability in a real-world simulation case study.

Main Methods:

• Developed a statistical model of queueing processes, focusing on patient departure rates and probabilities.
• Created a data-driven queueing network model using data from a major emergency department.
• Validated and applied the model using a synchronous simulation algorithm.

Main Results:

• The model accurately captures the complex interplay of patient arrivals and healthcare staff allocation (doctors, nurses).
• It provides an unbiased and precise measure of long-term emergency department crowding.
• Quantified the impact of new Unscheduled Care Services on crowding levels.

Conclusions:

• The novel data-driven queueing methodology effectively models and quantifies complex crowding dynamics in emergency departments.
• This approach successfully bridges the modeling gap, offering a tool to predict system crowding under various operational variables.