Recently, the food nutrition and functional component utilization research team develop the model prediction of herbicide residues in soybean oil that can be predicted using chemical analysis combined with mathematical models including MLR and PCA. The relevant research results were published in the journal Food Chemistry (IF: 7.514).

Pesticide residues in soybean products could accumulate and change during soybean processing, however, the pesticides depended on their physicochemical parameters and various processing methods. A nonlinear curve fitting equation of the form Y = aX² + bX + c and an MLR equation of the form Y = β₀ + β₁X + β₂X + β₃X were used. PCA has been extensively applied to reduce variable sets and extract a small number of potential factors to analyze the relationship between the observed variables.

This multivariate analytical method was used to investigate the correlation between the PFs of herbicides in the soybean oil and various physicochemical properties of the herbicides, including pKow, pSw, pMP, pVP, and pKa, the developed multiple linear regression model gave a fitting accuracy of ≥0.80 for predicting the theoretical PF values of pesticides in soybean oil products (0.39 < RMSE < 0.58). Thus, this model may be applicable for safety risk assessments and establishing maximum residue limits for pesticides in processed products.

Fig 4. Measured pPF values, as calculated using equations (1, 2), for 27 herbicides versus predicted pPF values: data from equations (1-1, 1-2) (A, B) and data from equations (2-1, 2-2) (C, D). Measured pPF values for EFSA oil samples, as calculated using equation (1, 2), for 19 pesticides versus predicted pPF values: data from equations (1-1, 1-2) (E, F) and data from equations (2-1, 2-2) (G, H).

Master student Jia Zhang and associate researcher Minmin Li as co-first authors, researcher Fengzhong Wang as corresponding authors. This study was supported by the National Key Research and Development Program of China (2017YFC1600600).

 

Link to the paper: https://doi.org/10.1016/j.foodchem.2021.131363