Knowledge translation becomes an important issue when mathematical modellers conduct collaborative research with public health researchers. Even for experienced researchers, nuances can get lost in translation. A case in point: disease incidence is an essential concept for infectious diseases. In public health terms, this is commonly defined as number of new cases over a period time, while in mathematical modelling, the same concept is often defined as number of new infections over a period. For infections that can be asymptomatic or subclinical, such as influenza, there is a huge difference between number of new cases and number of new infections over the same period.
A lesson can be learned here is that mathematical models typically predicts infections while surveillance data usually reports cases. When models are confronted with data, both modellers and public health researcher need to be aware of subtle differences among our terminologies.

I will show using simple models how differences between infections and cases would naturally lead to nonidentifiability when data is used to estimate parameter values by fitting models to data.