Abstract #M37

# M37
Validation of an automated body condition scoring camera.
Israel Mullins*1, Carissa Truman1, Jeffrey Bewley2, Joao Costa1, 1Dairy Science Program, Department of Animal and Food Sciences, University of Kentucky, Lexington, KY, 2CowFocused Housing, Bardstown, KY.

Body condition scoring (BCS) is the management practice of visually estimating subcutaneous fat reserves in dairy cattle. High and low BCS can negatively affect milk production, disease, and reproduction. Manually scoring BCS has proven beneficial yet can be difficult to implement. The desirable range for dairy cows varies and should be monitored at multiple time points throughout lactation for the most impact. A commercial automatic BCS technology is available for dairy cattle (DeLaval International AB, Tumba, Sweden). The objective of this study was to evaluate the agreement of the automated scores in comparison with conventional manual scoring. The study was conducted on a commercial farm in Indiana in April 2017. Three trained researchers scored cows manually, using a 1 to 5 BCS scale, with 0.25 increments. All data analyses were performed using SAS 9.3 (SAS Institute Inc., Cary, NC). Pearson correlation was calculated to assess interobserver reliability, the correlations being 0.85, 0.87, and 0.86. The automated scores were compared with 2 data sets of manual scores. One data set (MAN1) consisted of cows with ≥ 2 manual scores averaged by cow into one score per cow (n = 343). The second data set (MAN2) included cows that ≥ 2 scorers agreed on the score (n = 237). The mean was 3.38 ± 0.48 and 3.38 ± 0.78 (mean ± SD) for MAN1 and MAN2, respectively. The average automated score was 3.27 ± 0.27. Correlations were calculated for MAN1 and MAN2. Resulting coefficients were 0.78 and 0.77 (P < 0.001). The results from the t-test demonstrated equivalence between automated and manual data sets (P < 0.001). When scores were separated into manual scores: low (<3.00), average (3.00 to 3.75), and high (>3.75), the low and high BCS categories were shown to not be equivalent (P < 0.001). A Bland-Altman plot was constructed and demonstrated that the technology tended to underscore cows as their manual score increased. The system tends to be inaccurate at determining the extreme low and high BCS, although these categories represent a small proportion of cattle. An automated system may encourage more producers to adopt BCS into their practices.

Key Words: automatic measurement, technology, precision dairy farming