Skip to main content
ARS Home » Research » Publications at this Location » Publication #149092
Title: COMPARING CORRELATED PARAMETER ESTIMATES FOR NONLINEAR PET MODEL

Author
item WU, J - UNIV NEBRASKA
item PARKHURST, A - UNIV NEBRASKA
item ESKRIDGE, K - UNIV NEBRASKA
item TRAVNICEK, D - UNIV NEBRASKA
item Brown-Brandl, Tami
item Eigenberg, Roger
item HAHN, G - ARS COLLABORATOR
item Nienaber, John
item MADER, T - UNIV NEBRASKA
item SPIERS, D - UNIV MISSOURI

Submitted to: Meeting Abstract
Publication Type: Proceedings
Publication Acceptance Date: 4/1/2003
Publication Date: 4/27/2003
Citation: WU, J., PARKHURST, A., ESKRIDGE, K., TRAVNICEK, D., BROWN BRANDL, T.M., EIGENBERG, R.A., HAHN, G.L., NIENABER, J.A., MADER, T., SPIERS, D. COMPARING CORRELATED PARAMETER ESTIMATES FOR NONLINEAR PET MODEL. MEETING ABSTRACT. 2003.

Interpretive Summary:

Technical Abstract: The nonlinear PET model based on Newton's law of cooling can be used to estimate body temperature in cattle, Tb, challenged by hot cyclic chamber temperatures, Ta. The PET model has four biologically meaningful parameters: kappa, the thermal constant, delta, the difference between Tb and adjusted Ta; gamma, the proportion of variance in Tb comparable to variance in Ta; Tb initial, the body temperature at time zero. The two parameters gamma and delta are highly correlated in the current version of the model. This study looks at other ways to parameterize the PET model in an effort to reduce the correlation between parameters and improve nonlinear behaviors, such as parameter-effect curvature, bias, excess variance, and skewness.