Hybrid Modelling Approach Using Reinforcement Learning in Conjunction with Simulation: A Case Study of an Emergency Department

The last several years have seen a significant increase in adopting hybrid modellingHybrid modelling and simulationSimulation to model, understand, analyze, and enhance various healthcareHealthcare aspects, including patient flows, resource allocation, scheduling, policy evaluation, etc. The idea of hybrid simulationHybrid simulation is to develop simulationSimulation models combining at least two methodsDiscrete-Event Simulation: discrete-event simulationSimulation, system dynamicsSystem dynamics, andAgent-Based Simulation agent-based simulationSimulation. In contrast, hybrid modellingHybrid modelling combines various simulationSimulation approaches with modellingModelling methods other than simulationSimulation from the broader operations research and management sciences disciplines. The ability of hybrid modellingHybrid modelling and simulationSimulation to capture various intricacies and represent complex systems has encouraged researchers and practitioners to apply these techniques in healthcareHealthcare. While their application in healthcareHealthcare has increased exponentially, there are further opportunities to integrate advanced modellingModelling approaches. This chapter discusses a case study of an emergency department (ED) using a hybrid modellingHybrid modelling and simulationSimulation approach combining a forecasting, hybrid simulationHybrid simulation, and mixed-integer linear programming model to improve physician shift scheduling, patient flow and safety. Compared to the current practices, the proposed model reduced handoffs and patient time in the ED by 5.6% and 9.2% without a significant increase in budget. Finally, we propose our future work for integrating reinforcement learningReinforcement learning to further enhance the model.