This certified methodology presents an innovative, non-invasive, and cost-effective tool for the early prediction of metabolic and reproductive disorder risks in dairy cows based on milk composition analysis. The core of the proposed procedure lies in the application of Fourier-transform infrared spectroscopy (FTIR) to determine the detailed profile of fatty acids (FA) in milk fat, which serve as sensitive indicators of negative energy balance (NEB) and intensive lipomobilization.
The methodology describes in detail the relationships between the metabolic status of the animal and changes in fatty acid proportions, specifically the increase in FAs released from body reserves (C18:0, C18:1) and the decrease in FAs synthesized de novo in the mammary gland (C14:0). Utilizing advanced statistical methods, specifically binary logistic regression and ROC analysis, robust prediction models were developed separately for primiparous and multiparous cows. These models achieve high diagnostic reliability, with Area Under the Curve (AUC) values ranging from 0.81 to 0.90. The practical output of the methodology includes specific prediction equations and established cut-off values, enabling farmers and veterinarians to identify at-risk individuals with high sensitivity and specificity (up to 82%). Integrating these algorithms into milk recording systems allows for automated herd health screening, timely preventive intervention, and a subsequent increase in the profitability of milk production while improving animal welfare.