Co-Authors:
Maltz, E., ARO, Volcani Center, Institute of Agricultural Engineering, P.O. Box 6, Bet Dagan 50250, Israel
Antler, A., ARO, Volcani Center, Institute of Agricultural Engineering, P.O. Box 6, Bet Dagan 50250, Israel
Abstract:
A basic algorithm, based on once-daily observations of activity and lying behaviour of 15 cows that calved, was applied to data obtained from 12 dry cows before calving. These cows were kept in a dry-cow barn equipped to enable automatic downloading of behavioural data collected by a behaviour sensor fitted to each cow. On average the cows' steady behaviour changed significantly within 24 h prior to calving, with increases in daily steps and restlessness, and decreased lying time. The prediction of individual calvings was improved by incorporation of qualitative limits into the algorithm: 10 of the 12 calvings could be detected a day in advance, and 9 false positive alarms were reduced to 4 by incorporation of a restlessness variable, i.e., less than 5% of all the cow-days measured.