Noise Models in Respiration

The physiologically based fluctuations observed in experiments involving the human ventilatory response to inhaled carbon dioxide are incorporated in a number of stochastic models. Least squares is compared with maximum likelihood methods for different noise models. Multiple local maxima observed in the maximum likelihood technique motivate the development of a new method to detect the global maximum. Computation of parameter variances by a modification of the bootstrap method is developed as a check on the Cramer-Rao lower bound presently used.