Prediction of blood parameters of buffaloes from the milk mid-infrared spectra

Blood metabolic profile testing allows for monitoring metabolic health and nutritional status. Still, conducting extensive blood analyses on a large scale is not feasible due to high costs, labour, and animals’ stress. In this context, utilizing Fourier-transform mid-infrared (FT-MIR) spectra of milk to predict blood traits may present an effective opportunity. The present study aims to test the ability of milk FT-MIR to predict blood traits using milk spectra collected from the same buffalo. Blood and milk samples were collected from 310 buffaloes in different stage of lactation reared in 9 farms in the Southern Italy. The concentration of non-esterified fatty acids (NEFA), triglycerides, cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), glucose, urea, total protein, albumin, globulins, creatinine, aspartate aminotransferase (AST), alkaline phosphatase (ALP), glutamic oxaloacetic transaminase (GOT), glutamic pyruvic transaminase (GPT), gamma-glutamyl transferase (GGT), cre atine kinase (CK), lactate dehydrogenase (LDH) and total bilirubin (TBIL) were measured in blood through ref erence analysis. Milk samples were scanned with a MilkoScan and FT-MIR spectra were collected. Spectral data were divided into a calibration set (70%) and a validation set (30%) for external validation. In external validation, coefficients of determination ≥ 0.60 were achieved for urea, triglycerides, NEFA, LDL and TBIL concentrations. Although FT-MIR models cannot be considered accurate enough for punctual determination, the predictions of blood parameters can be considered for the screening of the herd and for genetic purposes at population level.