ASAS-NANP symposium: mathematical modeling in animal nutrition: overview of poultry nutrition modeling

Mathematical modeling has been used in poultry nutrition for the past 5 decades. Current amino acid recommendations for poultry have been based on mathematical models. This review aims to underscore the potential of modeling methodologies to minimize issues observed with the common use of empirical research, which researchers now realize. The review also discusses critical modeling issues and challenges to expanding modeling research. A comprehensive, although not exhaustive, list of existing models is presented to provide an overview of the efforts to develop these tools. Mechanistic models developed by EFG software and AVINESP are described in general terms since they have been well-documented over the past 3 decades. The framework and supporting data of these models are very similar. However, they differ in the research methodologies, including their parameterization and description of biological processes. The general methodology for model development and the fundamental equations are explained, and the current gaps in knowledge are discussed. The same concepts and description of growth, egg production, tissue and egg composition, and estimation of feed intake can be used to estimate needs for other nutrients and other animal species. The initial developments modeling poultry mineral nutrition are mentioned. Issues related to the accuracy and precision of these models might be resolved using big data, electronic sensors, portable devices to determine body composition, system-wide multi-omics, stable isotope technology, and machine learning techniques. Several publications have already demonstrated the practicality of integrating these methodologies. This review aims to demonstrate the relevance, applications, and solid basis of current mechanistic models that can be applied to advance sustainable poultry nutrition research. Modeling in poultry nutrition can help overcome many limitations observed using empirical methods and provide necessary decision-making tools. Models can be integrated with optimizers and feed formulation software.