Adsa Logo White Adsa Title White

Utilizing an ear-mounted accelerometer to estimate dry matter intake in transition dairy cows.

G. Mazon



Join G. Mazon on this page for a live text chat!
6:00 PM - 8:00 PM GMT


Utilizing an ear-mounted accelerometer to estimate dry matter intake in transition dairy cows.
G. Mazon*, M. R. Campler, J. H. C. Costa. University of Kentucky Lexington, KY.

The objective of the current study was to evaluate the use of precision dairy monitoring technologies to estimate individual DMI in transition dairy cows. Holstein dairy cows (n = 35; 1.9 � 1.1 lact.; 715.8 � 115.6 kg) were enrolled in the study from −28 � 3 d until 21 d post-calving. Cows were fed using intake feeders (Insentec, Marknesse, the Netherlands) that recorded individual DMI and fitted with ear tags (CowManager, Sensoor, Harmelen, the Netherlands). Cows were randomly divided into 2 data sets: training (n = 12) and validation (n = 23). The training data set was used to develop an algorithm to predict DMI based on mixed linear prediction models during the dry and fresh period. Prediction algorithms for the dry period included lactation number, BW, ear temperature, time active and inactive, and rumination time. The algorithm for the fresh period included lactation week, ear temperature, feeding, rumination, active and high active time. Predicted DMI was compared with actual DMI using Pearson correlation, linear regression, and Bland-Altman plots. The model was considered precise if the correlation coefficient and coefficient of determination were high (>0.70), and mean bias (predicted — observed DMI) from the Bland-Altman plot included zero within the 95% interval of agreement. The model was considered accurate if the slope from the linear regressions did not differ significantly from 1. Pearson correlation coefficients were moderate during the fresh period and low during the dry period (r = 0.55 and r = 0.39, respectively). The coefficients of determination were negligible (fresh R2 = 0.30; dry R2 = 0.13). Bland-Altman plots were acceptable with the mean bias � sd being −0.78 � 4.16 and 0.60 � 3.1 during the fresh and dry periods, respectively. The slope of the predictive model was 0.70 [95% CI; 0.41—0.99] during the fresh period and 0.32 [95% CI; 0.06—0.58]. In summary, DMI prediction models for the fresh and dry periods were not deemed precise or accurate. Future research should consider machine learning when developing DMI predicting models for transition dairy cows.

Keywords: automation, modeling, precision dairy.

Biography: Originally from Belo Horizonte, Brazil, Gustavo is currently a 1st year PhD student at the University of Kentucky under Dr. Joao Costa. Gustavo's PhD is focused on nutritional management strategies using precision dairy technologies. His career goal is to obtain a research and extension focused position at a land grant university.