Nutritional Decision-Making in Smart Farms: Interfaces between Animal Physiology and Algorithms

Background: Livestock industry globally faces a two-fold challenge in terms of increasing production output to meet the growing demand and at the same time reduce its effects on the environment. Classical average-based nutritional models themselves are ineffective, and they create resource wastage and inefficient animal performance. Aims: The current paper critically reviews the latest literature on integrating artificial intelligence (AI) and sensor technologies in the context of precision agriculture, especially making the model help in precision nutrition decision-making. Special emphasis is put on the shift of the concept of strictly algorithmic paradigms towards the systems that include real-time physiological measurements and modeling of physiologic metabolism. Methodology: The review of peer-reviewed articles with an impact is based on eighty articles that were selected through systematic literature review requirements and analyzed to reveal general tendencies and the methodological approaches. Results: The findings outline wearable biosensors and computer-vision systems to be one of the most promising sensor modalities that can provide high-resolution streams of temporal data. The software machine-learning and deep-learning techniques then convert such complicated datasets into predictive models that approximate the animal nutrient needs and physiological condition of a specific animal. In addition, a combination of these predictive algorithms and dynamic metabolic models offers a solid platform on which to produce a genuinely species-specific, that is, cow- or pig-species-specific, feeding regime. Conclusions: The implementation of the smart farm technologies promotes the development of precise animal nutrition, which promotes synergetic relationships between advanced algorithms and animal physiology.