Chemometric approach to characterizing and comparing the quality of buffalo meat from Nakhon Phanom and Khammouane provinces

Abstract

The results indicated that a chemometric approach could effectively characterize different attributes in quality between buffalo meat from Nakhon Phanom (NP) province, Thailand and Khammouane (KM) province, Laos. Neither the unsupervised principal component analysis (PCA) model nor the supervised partial least squares-discriminant analysis (PLS-DA) model completely separated the NP and KM groups. However, the sparse PLS-DA model was able to successfully distinguish between the meat samples originating from KM versus NP. Interestingly, orthogonal projections to latent structures discriminant analysis (OPLS-DA) exhibited superior discriminatory performances between regional meat samples. The robust OPLS-DA model used an orthogonal and a predictive factor, demonstrating a strong fit with R2X = 0.715, R2Y = 0.877 (P<0.001), and Q2Y = 0.803 (P<0.001). Consequently, two crucial variables were identified based on the selection criteria (VIP>2, P<0.05, FDR<0.05). Meat odors from sensors 1 (AUC=0.936, 95% CI: 0.841-0.989) and 4 (AUC=0.948, 95% CI: 0.843-1.000) could effectively distinguish between the NP and KM meats. In conclusion, the chemometric analysis successfully discerned regional quality differences and identified key discriminatory variables.