[en] Mineral contents in milk influence the humanand animal health. The variation of sodium (Na) could also be a good indicator of mastitis and are related to the milk fever. A low intake of calcium (Ca) promotes the osteoporosis in human. Consequently, knowingregularly the mineral contents in milk could be interesting. Unfortunately, the commonchemical analyses used to measure theconcentrations of mineral inmilk areexpensive.Theaimof this study was todevelopsome calibration equations to predict the contents of mineral directly in bovine milk using mid-infrared (MIR) spectrometry. Using a principal components approach, 70 samples were chosen based on their spectral variability from the 1609 collected milk samples. A total of 14 samples were considered as outliers. So, 57 samples were analyzed by inductively coupled plasma atomic emission spectrometry (ICP-AES) to estimate the reference concentrations of minerals. Based onthese values and the corresponding spectra, 3 calibration equations were built permitting to predict the contents of Ca, Na and phosphorus (P). To assess the accuracy of the developed equations, a full cross-validation was applied. In the same way, the RPD ratio corresponding to the ratio of the standard deviation to the standard error of cross-validation was estimated. The best accuracy was observed for the calibration equations estimating the contents of Ca (R2cv 0.82, RPD 2.35) and P (R2cv 0.77, RPD 2.06). In conclusion, the MIR spectrometry seemed to be a good alternative method to predict the contents of Ca and Pin bovine milk.
