In theory, we refine models until they perfectly represent reality.
In practice, that day never comes.
Every infectious disease model balances:
– Biological realism
– Data availability
– Computational feasibility
– Policy timelines
There is always another sensitivity analysis to run.
Another parameter to refine.
Another structural assumption to debate.
The real question is not whether a model is perfect.
It is whether it is useful.
A model may be “good enough” when:
• It captures the dominant transmission mechanisms
• Its uncertainty is transparently presented
• Its assumptions are clearly stated
• Its limitations are openly discussed
• It informs a decision that cannot wait
Overfitting can be as misleading as oversimplification.
A complex model with fragile assumptions is not automatically superior to a transparent, robust framework.
The pressure in applied epidemiology is real:
Decision-makers often need answers before the science feels complete.
That tension is not a weakness of modeling.
It is part of its responsibility.
The goal is informed judgment under uncertainty.