Accelerometers-based position and time interval comparisons for predicting the behaviors of young bulls housed in a feedlot system

• DOI: 10.1016/j.atech.2024.100542
• Publication Date: 2024-08-23T22:49:00Z
• Authors: Rafael Nakamura W...; Eliéder Prates Ro...; Priscila Arrigucc...; Julia Lisboa Rodr...; Guilherme Alves d...; Matheus Mello Silva; Márcia Helena Mac...; Sabrina Luzia Cae...; Salvador Boccalet...; Ricardo Andrade Reis; Danísio Prado Munari
• Keywords: Feedlot; Position (finance); Body position

Animal behavior monitoring is an important tool for animal production. This strategy can indicate the well-being and health of animals, which lead to better productive performance. study aimed assess most effective accelerometer attachment position (on either halter or a neck collar) data transmission time intervals (ranging from 6 600 s) predicting behavioral patterns, including water food intake frequencies, as well other activities in young beef cattle bulls within feedlot system. A range machine learning algorithms were applied satisfy aims study, random forest, support vector machine, multilayer perceptron, naive Bayes classifier algorithms. All studied models produced high performance metrics (above 0.90) when using both positions, except built classifier. Therefore, coupling accelerometers with collars more viable alternative use on doing so easier than applying halters. Utilizing dataset observations (i.e., shorter intervals) did not result considerable improvements trained models. datasets fewer advantageous, it decreased computational temporal demands model training, addition saving battery device considered this study.