Use of MIR spectral data of milk in the detection and prevention of lameness in dairy cows

Currently, there is no prediction equation that enables the detection or prediction of lameness problems in dairy cows based on milk composition. Lameness is an important health issue for the animal and economic issue for the farmer. Therefore the general objective of this study was to test the feasibility of detecting lameness problems using mid-infrared (MIR) spectra from milk, as this could provide an easy and early detection method for the farmer. The data originated in Austria and was therefore distributed across three breeds; Austrian Simmental, Brown Swiss and Holstein. Locomotion scores, levels 1 (sound) to 5 (severely lame), given to the animals were organized into a classification variable non-lame or lame with a threshold of 2. Two different MIR calibration methods were tried with the aim of obtaining a reliable prediction equation for lameness. The first method was classic MIR calibration and used only the MIR spectra as predictive variables. In this case, trying to obtain better results was achieved through selection of subsets in such a way that overall variability would be reduced. The second method used extra predictive variables in the form of milk biomarkers for lameness to aim for a better prediction. The resulting sensitivities and specificities for both methods very rarely went above 80% and mostly seemed to stagnate around 60 to 70%. These results are certainly not high enough for application in the field, but this study does suggest the existence of a real, if complex, link between milk composition and lameness, through foot and claw lesions and metabolic disorders, which leaves the door open for further studies.