| Sumario: | This study focuses on the application of new technologies in the livestock sector, particularly
those aiming at improving individual animal management through data collection and analysis
of animal behavior. The goal is to use this data as livestock health and well-being indicators.
The study incorporates a commercial triaxial accelerometer called Digitanimal® into goat
farms to provide information about individual animal behavior. The study was conducted at
an experimental farm for small ruminants, using a herd of 14 Murciano-Granadina goats in a
non-productive state. The accelerometers were integrated into collars worn by the goats,
recording position values of the X, Y, and Z axes at a frequency of 10 Hz. Behaviors such as
lying down, standing, rumination, fighting, movement, and inactivity were observed and
recorded during 4 hours per day, always in the morning. Behaviours were recorded through
an app provided by Digitanimal®, specially designed for the validation of the devices. The
collected data were synchronized with the recorded behaviors, and the predictive capacity of
the accelerometer was evaluated using the Random Forest machine learning algorithm. The
results showed that the Random Forest algorithm had a moderate to high prediction capability
for behaviors such as inactive and lying down, with 46% and 100% accuracy, respectively.
The algorithm also performed well in identifying movement behavior, with an accuracy of
82%. However, rumination and fighting behaviors could not be evaluated due to the limited
amount of data collected.
The study demonstrates that this sensor could predict behaviors such as active, inactive,
movement, and lying down in goats. However, more data and observation hours are needed
to improve the accuracy of predicting rumination and fighting behaviors. The accuracy of
human observers in labeling behaviors is also an important factor that influences data
synchronization, training, and validation. In conclusion, this study highlights the potential of
using these specific accelerometers and machine learning algorithms to monitor and assess
livestock behavior, providing valuable and promising insights into animal health and well being
|
|---|