Data assimilation in emergency department simulations for real-time disaster response

Mass casualty incidents (MCIs) due to disasters such as earthquakes and infectious disease outbreaks create a sudden surge of patients presenting to the emergency department (ED). Responding appropriately to these incidents requires EDs to have situational awareness of both the current and future impacts of an MCI on patient presentation and resource allocation. ED simulation models are used to support such decision making by simulating how MCIs will affect an ED based on hypothetical assumptions or historical data from past MCIs. However, because each MCI can unfold differently, there remains a need to update these ED simulation models using ED observations collected as an MCI unfolds to adequately represent the ED’s current conditions in reality. Here, we present a novel data assimilation method for incorporating recently collected observations into an ED simulation model to align the conditions in the simulation model with the conditions in its real-world ED counterpart during an unfolding MCI. We demonstrate that using real-time ED observations can generate more accurate and reliable estimates of the ED’s current conditions and forecasts of its future conditions, compared with relying solely on past observations. Our results highlight how ED simulation models can function as real-time decision support tools in the ED, particularly for enhancing situational awareness during MCIs to support real-time disaster response. • Data assimilation allows emergency department (ED) simulations to use real-time data. • Using real-time data improves model estimates of current and future ED conditions. • ED simulations can enhance situational awareness during mass casualty incidents.