Modeling and analysis of synergistic phenolic antioxidants in edible oil: Integrating molecular simulation and artificial neural network approach

Abstract

Oxidation causes edible oil deterioration affecting nutrition and safety. Antioxidants delay aging, but selecting compounds experimentally is time-consuming and arbitrary. Thus, a rapid and rational screening method for synergistic antioxidants is essential. This study evaluates the antioxidant effects and mechanisms of four phenolic antioxidants on the oil matrix using multiscale simulation. Tert-butylhydroquinone (TBHQ) exhibited the lowest dissociation energy (226.92 kJ/mol), followed by flavanol (TP) (233.00 kJ/mol). Syringic acid (CA) showed the smallest mean square displacement value. An artificial neural network (ANN) was established to analyze the synergistic and antagonistic interactions between TBHQ and TP, predicting optimal synergy at 3.6 mg TP and 1.8 mg TBHQ. The relative prediction deviation (RPD) value of the validation set was 3.91, and the relative error production (RER) value was 1.87 %. Fourier transform infrared spectroscopy experiments validated this synergy. This research establishes a screening system for compound antioxidants and models the synergistic effects of phenolic antioxidants in edible oil, offering theoretical and technical support for their application in the food industry.