Comparison of two software methods of fitting one- and two-compartment age-dependent digesta kinetics models

Authors: Moffet, Corey; Gunter, Stacey

Submitted to: American Society of Animal Science Southern Section Meeting
Publication Type: Abstract Only
Publication Acceptance Date: 11/8/2018
Publication Date: 7/29/2019
Citation: Moffet, C., Gunter, S.A. 2019. Comparison of two software methods of fitting one- and two-compartment age-dependent digesta kinetics models. Journal of Animal Science. 97(E Supplement 1):65-66. https://doi.org/10.1093/jas/skz053.147.
DOI: https://doi.org/10.1093/jas/skz053.147

Interpretive Summary:

Technical Abstract: Ruminant nutritionists fit digesta kinetics models to better understand processes in the gastrointestinal tract. Several software packages are available to fit one- (G2) and two-compartment (G2G1), age-dependent digesta kinetics models to fecal marker concentration over time after giving a marker dose. We characterized repeatability and agreement for two programs (R and SAS) used to fit these models. We constructed 8100 datasets using a Monte-Carlo approach where errors were added to true marker concentrations for samples from 81 synthetic profiles representing different animal and diet combinations (100 datasets per profile). Datasets contained concentrations at 15 nominal times (from 0 to 120 hours after dosing). The two error sources were sampling time (uniform between -3 and 1 hours) and concentration measurement (normal with mean 0 and SD = 5 + 0.08CT, where CT is true concentration). The resulting datasets were fit to G2 and G2G1 models using both software and model parameters (K0, ' or '1, K2, and '') were used to calculate digesta kinetics parameters (particle passage rate, gastrointestinal DM fill, fecal DM output, gastrointestinal mean retention time, and rumen retention time). When fitting G2 models, all converged, but when fitting G2G1 model, 60 did not converge. The parameters were more repeatable from R than from SAS, but the ratios of repeatability to the mean was generally less than 10%. Bias and SD of differences between software packages were also small. The G2G1 models produced smaller bias and SD of differences than the G2 models. Bias and SD for digestion parameters between software were also small, however G2G1 models had smaller biases and SD of differences than G2 models. Repeatability was better with R than with SAS, but mean differences were small; the G2G1 model was more repeatable than the G2 model, but differences between software were small for G2G1 models.