MACHINE LEARNING IN ANIMAL NUTRITION: PREDICTING FEED EFFICIENCY, GROWTH AND NUTRIENT UTILIZATION

Resumen

Machine learning (ML) has expanded the analytical capacity of animal nutrition by enabling the integration of animal, dietary, environmental and production data. This review aimed to examine current applications of ML in predicting feed efficiency, growth and nutrient utilization in livestock, focusing on data sources, modelling approaches, performance and practical limitations. A narrative literature review was conducted through a structured search of scientific databases, prioritizing peer-reviewed studies published between 2021 and 2026, with selected earlier references used for conceptual and methodological background. The reviewed studies show applications of ML in the prediction of dry matter intake, feed efficiency, nutrient requirements, energy partition, growth and digestibility using phenotypic, dietary, behavioural, sensor-derived, genomic and microbiome data. Random Forest, gradient boosting, support vector methods and artificial neural networks have been used to capture nonlinear relationships that are difficult to represent through conventional nutritional equations. However, small datasets, heterogeneous conditions, missing data, overfitting and limited external validation restrict model transferability across breeds and production systems. Recent approaches increasingly combine explainable ML, mechanistic knowledge and sensor technologies. ML therefore offers potential for advancing individualized nutritional management and supporting precision feeding, provided that predictive models are validated across diverse production conditions.