Application of Organic Acid Based Artificial Neural Network Modeling for Assessment of Commercial Vinegar Authenticity

Abstract

Vinegar as a nutraceutical substance is classified to various types related to the different substances applied in production process. Therefore, identity of the source and authenticity of the samples would be inevitable. The present study addresses determination of organic acid composition of 47 vinegar samples categorized to four types including distilled, apple, grape, and pomegranate and uncategorized vinegars (5, 12, 15, 3, and 12 samples, respectively). High-performance liquid chromatography (HPLC) method was performed according to ICH guidelines using simple sample preparation for determination of eight organic acids including oxalic, formic, ascorbic, lactic, acetic, malic, citric, and propionic acids. Findings were treated by a nonlinear computational analysis called artificial neural network (ANN) utilizing a back propagation method for training the multilayer feed-forward neural network to determine vinegar type. HPLC method resulted in suitable separation where limit of detection (LOD) and limit of quantification (LOQ) were ranged from 0.11 to 0.89 ppm and 0.34 to 2.69 in malic acid and oxalic acid, respectively. The recovery process was also ranged from 97.1 to 106.4 for oxalic acid in apple vinegar and lactic acid in grape vinegar, respectively. ANN modeling indicated a comparative model to recognize sample origin where accuracy estimation was 88.6 %. The obtained model was applied to determine the probable origin of some uncategorized commercial vinegars. It was concluded that ANN model along with analytical methods such as HPLC could be established for evaluation of commercial samples in food control laboratories.