Summary:
Monitoring blood metabolites is essential for assessing dairy cattle health, but traditional methods are time-consuming, costly, and stressful for cows. This study explored a more efficient approach by integrating in-line near-infrared (NIR) milk spectroscopy with on-farm data (days in milk and parity) and genetic markers to predict blood metabolite levels in Holstein cattle.
Researchers analyzed data from 388 Holstein cows using an AfiLab system, applying three models:
- Model 1 (M1): Used only NIR milk spectra.
- Model 2 (M2): Combined NIR with on-farm data.
- Model 3 (M3): Incorporated NIR, on-farm data, and genetic markers.
Key Findings:
- Adding on-farm data (M2) improved prediction accuracy by:
- 19% for energy-related metabolites (e.g., glucose, cholesterol)
- 20% for liver function markers
- 7% for inflammation markers
- 24% for oxidative stress metabolites
- 23% for mineral levels
- Including genetic markers (M3) further enhanced predictions:
- 34% for energy-related metabolites
- 32% for liver function markers
- 22% for inflammation markers
- 42% for oxidative stress metabolites
- 41% for mineral levels
- Selecting specific genetic markers through genome-wide association studies (GWAS) improved accuracy further, though with minor reductions in phosphorus, glucose, and antioxidant power prediction.
Conclusion:
This study demonstrates that integrating NIR milk spectra with on-farm and genetic data significantly enhances the accuracy of blood metabolite predictions. This approach could enable large-scale, real-time health monitoring in commercial dairy herds, reducing costs and improving animal welfare.
