A smart based screening system by MicroNIR and chemometrics for on-site authentication of buffalo milk in dairy industry

Abstract

Buffalo milk represents one of the most interested dairy products involved in adulteration practice, as the current yield does not satisfy the increasing demand of the market. The development of an analytical system able to identify adulteration, defending manufacturers/retailers as well as consumers, represents an important challenge for the entire scientific community and national authorities involved in controls. In this study, an analytical system based on an easy-to-use device and chemometric tools was proposed for a rapid screening of the raw material, the buffalo milk. Especially, a spectroscopic method was optimized for the analysis of pure raw material and buffalo milk after adulteration with goat milk, cow milk and water. Spectra were studied by techniques of multivariate statistical analysis. After an explorative investigation of the spectroscopic results, prediction models were validated. The Partial Least Squares-Discriminant Analysis (PLS-DA) model provided accuracy higher 93.7 % and the Soft Modeling Class Analogy (SIMCA) model showed a sensitivity never lower than 91.3 %. The Partial Least Squares regression (PLSr) model ensured a rapid assessment of contamination, providing an error of prediction (RMSEP) never higher than 5.2 %. The proposed MicroNIR/Chemometric system proved to be a rapid and sensitive tool for real-time investigation of dairy products at any farm levels.