The use of animal sensor data for predicting sheep metabolisable energy intake using machine learning

The use of sensors for monitoring livestock has opened up new possibilities the management in extensive grazing systems. work presented this paper aimed to develop a model predicting metabolisable energy intake (MEI) sheep by using temperature, pitch angle, roll distance, speed, and time data obtained directly from wearable on sheep. A Deep Belief Network (DBN) algorithm was used predict MEI, which our knowledge, not been attempted previously. results demonstrated that DBN method could MEI sensor alone. mean square error (MSE) values 4.46 20.65 have achieved training testing datasets, respectively. We also evaluated influential variables, i.e., distance MEI. Our study demonstrates application machine learning techniques on-animal presents substantial opportunity interpret biological interactions systems data. expect further development refinement technology will catalyse step-change management, as become widely producers.