Use of automated systems for recording of direct and indirect data with special emphasis on the use of MIR milk spectra (OptiMIR project)

A current tendency in developing tools to support farm management is to make use of advanced sensors closely associated to animals, facilitating the collection of large quantities of data ideally at a low cost without perturbing the animal itself. On a dairy farm level, sensors measuring milk conductivity or pedometers measuring mobility are often cited as examples. This introduces the concept of "precision livestock farming" where a given "bioresponse" captured by a "biosensor" allows the creation of feedback to adjust the "bioprocess". Such on-farm systems are often restricted to a given farm and they are mostly strictly separated from standard performance recording systems. In dairy cows, a particular rich source of information to detect a "bioresponse" is milk and its (fine) composition. Standard milk analysis undertaken in milk recording schemes by mid infrared spectroscopy (MIR) generates spectral data that reflects the milk characteristics. Therefore, spectral data directly reflects the metabolic (e.g., energy balance) and health (e.g., udder health) status of the cow. The use of MIR spectral data to predict fine milk components (e.g., fatty acids) is now becoming commonplace. However the use of MIR spectral data could provide an even more direct method to assess the "bioresponse" in relation to health, fertility, feeding, milk quality and even rejection of pollutants. For this reason, 12 EU milk recording organizations and milk laboratories together with 6 EU research groups have joined forces to develop the North-West Europe INTERREG IVB Project OptiMIR (www.optimir.eu). As a first step to use spectral data for developing decision support tools, the project includes the development of methods to standardize spectral data generated by various apparatuses in different laboratories. Through the OptiMIR project, health indicator traits from milk analysis either through the prediction of milk components (i.e. lactoferrin) or through the direct assessment of the health status of the cow (i.e. clinical mastitis) will become available. These data can then be generated in routine milk recording and can be stored in a central database. Because generating MIR data at the on-farm level is still difficult and expensive, the use of near infrared (NIR) spectroscopy is currently also under investigation by other groups. For a comprehensive use of fine milk composition, as for other automated sensors, the optimum would be a close and bi-directional interaction between in-line on-farm systems and central databases in order to contribute to the successful implementation of powerful health monitoring systems and decision support tools.