Livestock Growth: Machine Learning Review of the Literature

Abstract

Livestock growth prediction plays a critical role in farm management, animal welfare, and productivity optimisation. While numerous studies have explored the use of machine learning for modelling animal weight and growth, most rely on curated datasets collected under controlled conditions, with limited consideration for the variability and inconsistency of real world farm data. This narrative review synthesises current research on livestock growth prediction with a particular focus on data collected in pasture based systems. We examine the types of data available to producers ranging from animal characteristics and weight records to feed intake, environmental factors, and treatment logs and compare how these have been used in prediction models across cattle and sheep. The review highlights a dominance of research centred on cattle, with sheep comparatively under explored despite their global significance. Additionally, we identify key challenges in modelling growth using mob (group) records, where animals are managed in groups rather than individually. In response, we evaluate emerging machine learning techniques, including probabilistic approaches and ensemble models that offer potential for handling uncertainty, integrating multi modal data, and adapting to the complexity of real life production systems. This review concludes by outlining future directions for applying robust, scalable predictive models to improve livestock growth forecasting in pasture grazed systems.

Keywords: Machine learning, livestock, weight growth, data utilisation, sustainability