Prediction of energy status of dairy cows using MIR milk spectra
- Grelet, C. , Vanlierde, A. , Froidmont, E. , Gpluse Consortium, & Dehareng, F. (2017). Prediction of energy status of dairy cows using MIR milk spectra. Proceedings in: EAAP annual conference, Tallin, 28-08-2017, p 403.
|Year of conference||2017|
|Title||Prediction of energy status of dairy cows using MIR milk spectra|
|Conference name||EAAP annual conference|
|Volume||Book of abstracts|
|Abstract||A key task within the GplusE project is to undertake a genetic evaluation using ten thousand cows to improve health traits of dairy cows, with energy status (ES) being a trait of major interest. To achieve this, cost effective and large scale phenotyping methods are required. This study was designed to evaluate the possibility of using MIR spectra of milk to predict ES of cows. Data was collected from 241 cows, from calving until 50 days in milk (DIM) in six research herds of the GplusE consortium distributed in Belgium, Denmark, Germany, Ireland, Italy and UK. Milk MIR spectra were collected twice weekly during this period. ES was ’quantified’ by measuring daily energy balance (EB), residual feed intake (RFI), dry matter intake (DMI), and measuring at 14 and 35 DIM blood metabolites/hormones (Glucose, BHB, NEFA and IGF-1). K-means clustering was also performed based on these 4 blood components in order to discriminate 2 groups with healthy vs imbalanced cows. Regression models between each of these variables and MIR milk spectra have been developed using PLS and classification model with SVM method. The R² of cross-validation obtained when predicting EB, RFI, DMI, Glucose, BHB, NEFA and IGF-1 were respectively 0.43, 0.46, 0.47, 0.31, 0.40, 0.28 and 0.48. Discriminant model based on blood metabolite clusters was able to differentiate healthy vs imbalanced cows with sensitivity and specificity of 84% and 81%, respectively. These preliminary results demonstrate that milk MIR spectra have reasonable potential to provide information on ES related variables. This could allow large scale predictions for both genetic studies and farm management.|
|Authors||Grelet, C., Vanlierde, A., Froidmont, E., Gpluse Consortium, , Dehareng, F.|