Transmission infrared spectroscopy for rapid quantification of fat, protein, and lactose concentrations in human milk

To develop partial least squares regression (PLSR) calibration models in combination with transmission infrared (TIR) spectroscopy for rapid and optimal quantification of human milk macronutrient concentrations. Human milk samples (n = 306) were characterized simultaneously by reference chemical analytical methods and TIR spectroscopy. Reference macronutrient concentrations were linked to pre-processed spectra and divided into two (training and test) sets. PLSR was used to develop trial calibration models using training set, and the test set was used to assess the accuracy of the trial analytical methods. For the methods selected as optimal, the concordance correlation coefficients between reference and TIR-based methods were 0.93 for fat, 0.96 for protein, and 0.52 for lactose. The Bland-Altman plots showed no evidence of systematic bias between TIR and reference methods. TIR spectroscopy provides the basis for accurate and rapid quantification of human milk fat and protein concentrations but is less accurate for measuring lactose concentration.