Forcing function selection for identification in human respiratory control

This paper considers input forcing function selection for discriminating among alternative structural models and for estimating parameters of functional models. The model outputs are corrupted by an additive noise source, which represents the so-called breath-to-breath variation in the observation of a ventilation response to an end-tidal CO2-O2 forcing function input. The problem of forcing function selection for model discrimination is formulated around minimizing the probability of error in model selection while the problem of forcing function selection for model parameter estimation is formulated around minimizing the parameter estimation error variance. Forcing function selection is illustrated by considering end-tidal CO2 inputs that enhance parameter estimation for a functional model that summarizes normoxic respiratory controller behavior.