Predictive Models of Dairy Cow Thermal State: A Review from a Technological Perspective

Simple Summary Heat stress in cattle is broadly defined as a physiological condition in which body temperature rises, and the animals are no longer able to adequately dissipate body heat to maintain thermal equilibrium due to environmental factors. Dairy cattle are particularly sensitive to heat stress because of the higher metabolic rate needed for milk production. Due to global warming and the expected growth of milk production in warmer regions, an increase in the occurrence of heat stress can only be avoided with the use of environmental control systems. However, most available systems were developed to take corrective measures or are not accurate enough to effectively prevent heat stress, as there is not yet an automated technological solution that considers all the environmental and animal variables that determine the occurrence of this condition. Further, these systems must be connected in time to prevent this condition in cattle but also disconnected when they are no longer needed, as their use raises major economic and environmental concerns regarding energy and water consumption. This review describes and discusses three types of predictive models that can make these systems more effective in preventing heat stress and more efficient in the use of energy and water. Dairy cattle are particularly sensitive to heat stress due to the higher metabolic rate needed for milk production. In recent decades, global warming and the increase in dairy production in warmer countries have stimulated the development of a wide range of environmental control systems for dairy farms. Despite their proven effectiveness, the associated energy and water consumption can compromise the viability of dairy farms in many regions, due to the cost and scarcity of these resources. To make these systems more efficient, they should be activated in time to prevent thermal stress and switched off when that risk no longer exists, which must consider environmental variables as well as the variables of the animals themselves. Nowadays, there is a wide range of sensors and equipment that support farm routine procedures, and it is possible to measure several variables that, with the aid of algorithms based on predictive models, would allow anticipating animals' thermal states. This review summarizes three types of approaches as predictive models: bioclimatic indexes, machine learning, and mechanistic models. It also focuses on the application of the current knowledge as algorithms to be used in the management of diverse types of environmental control systems.