Traditional methods of monitoring health changes in animals are based entirely on the human senses. However, in modern dairy production systems humans are rarely present, this is particularly the case with the introduction of robotic milking. In these systems all the functions of milking are automated and cows visit at times of their own choosing. Systems of automatic health monitoring are therefore a priority of research to ensure that the health and reproductive starus of the animals can be assessed for management purposes. These systems must be automatic, work in the field conditions without technical support and cost a few pence per analysis. The first task is to obtain representative biological samples automatically and non-invasively. As milk is flowing into the milking machine from the cow this can be achieved with ease, except that milk is non-homogeneous with a changing lipid fraction during milking. Lipid soluble components such as progesterone and vitamin A are affected by this change and a model has to be established to determine thresholds at different times during milking. Our main interests in dairy cows are in predicting ovulation, detecting metabolic imbalance and detecting preclinical mastitis inflammatory response. Our team is developing a fully automated ovulation prediction system based on the screen-printed carbon electrode biosensor for progesterone demonstrated by Pemberton et al. (1998). In recent experiments the automated system was able to detect concentrations of progesterone between 2 and 30 ng/ml in stored milk samples (r2 = 0.96). The results of field tests are presented showing a good correlation between ELISA and the biosensor (r2 = 0.91) on samples of fresh milk. The results of the recent field tests show the ability of the biosensor to characterise ovulation cycles of cows and to detect pregnancy. We have identified a major lack of other biological models to detect disease with on-line sensors. Our next objective is to create an integrated system for biological research with sensor systems for urea, ketones, lipids and enzymes in milk. This will allow the development of diagnostic models based on analysing numerical sensor-derived data rather than human visual observations for signs of ill health in dairy cows.
Mottram, T.; Velasco-Garcia, M. ; Berry, P.; Richards, P.; Ghesquiere, J. and Masson, L. (2002). Automatic on-line analysis of milk constituents (urea, ketones, enzymes and hormones) using biosensors. Comparative Clinical Pathology, 11(1) pp. 50–58.