Validation of a multiple accelerometer sensor system to estimate dry matter intake in lactating dairy cows

Document Type

Abstract

Publication Date

2019

Location

2019 American Dairy Science Association Annual Meeting: Cincinnati, Ohio

Publisher

American Dairy Science Association

Journal

Journal of Dairy Science

Volume

102

Issue

Suppl.1

Pages

134

Language

en.

Keywords

accelerometer, intake, sensor technology

Abstract

Dry matter intake is a valuable parameter that can be used to evaluate health and milk efficiency in dairy cows. The objective was to evaluate the use of 3-dimensional accelerometer sensors to estimate DMI in lactating dairy cows. Twenty-four late-lactation Holstein dairy cows housed in a free-stall barn were fitted with 3 sensors that record acceleration in the 3-axis (i.e., x, y, and z), one sensor on the lateral side of the left hind leg and 2 attached to a halter directly superpose over the jaw and nose. Cows were assigned 2 groups, a data collection group (A; n = 12) and a validation group (B; n = 12). Cows were trained to use Calan gates during a 7-d period followed by 10 d of data collection of acceleration and individual intakes for both groups. Four cameras were used to continuously video record cows, then eating times for each cow were generated. Sensors were set to record the accelerations at 10-s intervals. Eating times and accelerometer data from group A was cross-reference based on date and time. Six new variables were derived from jaw and nose accelerations by measuring the change in acceleration between 2 consecutive time points (lag-time). The REG procedure of SAS was used to regress each acceleration against the DMI during 24h period in group A to select the highest R2 data. In group B, DMI derived from acceleration combinations (DMIA) was compared against the actual DMI using the CORR and MIXED procedures of SAS. All 32,767 acceleration combinations were tested, and only 921 were deemed relevant (>80% data coverage). LagNoseZ+LagNoseX model had the strongest positive correlation (r = 0.42), but its DMIA was greater (P < 0.01) than the actual DMI (36.2 vs 21.4 kg/d). In contrast, LegY+JawZ+LagNoseY model had similar DMIA and actual DMI (P = 0.99) but a weaker correlation (r = −0.11). The LegX+JawX+LagNoseZ model had a weak correlation (r = 0.28), but an accurate (P = 0.77) estimation of actual DMI (21.8 vs 21.4 kg/d). Accelerometer sensors have a great potential to estimate actual DMI in dairy cows, but a superior correlation will be required to improved robustness and reliability in a commercial farm.

Link to Full Text

Share

COinS