What if we could predict how many patients arriving at an emergency department would need hospital admission?

Would that be enough to prevent overcrowding?

Knowing what is likely to happen can help hospitals prepare. But anticipating demand is only part of the challenge.

In 2026, a study published in Emergencias by Millán Soria and colleagues explored the use of machine learning to predict hospital admissions from emergency departments in real time. Using data from more than 328,000 emergency department episodes, the researchers developed a predictive model and integrated its results into a dashboard to support inpatient bed planning.

This is an important step. Better information can help hospitals anticipate their needs and make more informed decisions.

But there is another question we need to ask: once we know what is coming, will the hospital be able to absorb it?

A prediction does not make an inpatient bed available. It does not shorten the time required to prepare a bed after discharge. Nor does it guarantee that a patient who has been admitted can leave the emergency department promptly.

These processes still depend on available resources, organisational decisions and the time required to move patients through the hospital.

Imagine that a hospital anticipates an increase in admissions tomorrow. That information is useful, but its practical value depends on what the organisation can do with it. Can discharges be coordinated? Will beds become available when they are needed? Can transfers take place without creating new delays?

The same predicted demand may produce very different operational outcomes depending on the answers.

This is where two complementary perspectives become relevant.

Predictive models help us anticipate what may happen. Simulation models allow us to explore how a system may behave under different conditions.

In my Master’s research in Hospital Engineering and Architecture, I developed the DFU simulation model to examine how changes in demand, patient characteristics, resource availability, process times and patient flow affect emergency department operational capacity.

The model does not predict future admissions or operate as a real-time hospital forecasting system. Its purpose is different: to explore scenarios and examine how changes in selected parameters affect the system’s behaviour.

Together, prediction and simulation address two distinct questions. What is likely to happen? And what might happen to the system when it does?

For hospital planners and managers, both questions matter. Anticipating demand is valuable, but planning also requires an understanding of the capacity available to respond to that demand.

A hospital needs to know what is coming. It also needs to know whether its resources and processes will be able to cope with it.