Computational tools for automatic weighing platforms aiming the detection of nutritional and sanitary events in beef cattle

Abstract

The conventional live weight monitoring system in beef cattle is characterized by the removal of the animals from the pens or paddocks and handling them in corrals with static weighing systems. In this process, several problems related to calibration of scales and handling of the squeeze chute, added to fluctuations in live weight that occurs throughout the day, mainly due to changes in the gastrointestinal tract and body fluids contents, decrease the accuracy of information collected. Furthermore, the extension of the intervals between measurements typically performed every 28 days or in even longer intervals, results on the absence of information that could be used for the intervention of nutritional and sanitary problems in time to avoid a reduction in animal performance. The speed of this intervention, with corrective attitudes, may result in economic benefits to the producer, decrease in the price of the commodity to the consumers and even in environmental benefits for society. In this context, the present study will evaluate the automatic body weighing system in a beef cattle feedlot, aiming to analyze its potential in detecting problems in the performance of the pen, as well as physiological disorders at initial stage in the individuals. For this, 250 Nelore cattle will be daily monitored in feedlot pens equipped with an automatic body weighing platform system (Intergado®, model VW 1000), which allows the collection of several weights of the same animal throughout the day without the stress associated with conventional weighing procedures and without affecting rates of gain. The data collected by this system will be processed and analyzed using a non-parametric approach to provide a smoothed performance curve of the animals, allowing monitoring of their real behavior. Moreover, the animals will also be monitored with cameras coming from old FAPESP approved projects (2012/03296-5), aiming the creation of a database, which may later be associated with variations in performance observed by the weighing platforms, allowing the development of machine learning subroutines. We hope that these methodologies that will be explored and tested in the present study will enable those sensors connected to computers to identify physiological problems or other conditions that affect animal performance. This research is fundamental for the future development of precision livestock farming for beef cattle. Ultimately, we are looking for innovative tools that may decrease costs, minimize the environmental impact and provide positive effects in terms of animal welfare through more rapid detection and identification of any anomaly. (AU)