Applications of Machine Learning in Livestock Production and Genetic Improvement

Abstract

Machine learning (ML) is increasingly being explored in livestock research because of its ability to analyse large and complex genomic, phenotypic, environmental and image-derived datasets. This review provides a concise overview of recent applications of ML in livestock, with emphasis on genomic prediction, genetic improvement, production prediction, high-throughput phenotyping and precision livestock farming. Recent studies indicate that approaches such as random forest, support vector machines, neural networks and deep learning can improve prediction or automate phenotypic measurements for selected traits and applications. However, ML does not consistently outperform established genomic prediction methods such as GBLUP and Bayesian approaches, with performance varying across traits, populations and datasets. Major challenges include limited and heterogeneous datasets, model generalizability, computational requirements and the interpretability of complex models. Future developments should focus on standardized datasets, explainable AI, independent validation and integration of genomic, phenotypic, environmental and multi-omics information. Overall, ML offers promising opportunities to complement conventional livestock breeding and support more efficient, data-driven genetic improvement and farm management.