Optimization of accelerated solvent extraction of ellagitannins in black raspberry seeds using artificial neural network coupled with genetic algorithm

Abstract

This study aimed to identify ellagitannins in black raspberry seeds (BRS) and to optimize accelerated solvent extraction of ellagitannins using an artificial neural network (ANN) coupled with genetic algorithm. Fifteen monomeric and dimeric ellagitannins were identified in BRS. For ANN modeling, extraction time, extraction temperature, and solvent concentration were set as input variables, and total ellagitannin content was set as output variable. The trained ANN had a mean squared error value of 0.0102 and a regression correlation coefficient of 0.9988. The predicted optimal extraction conditions for maximum total ellagitannin content were 63.7% acetone, 4.21 min, and 43.9 °C. The actual total ellagitannin content under the optimal extraction conditions was 13.4 ± 0.0 mg/g dry weight, and the prediction error was 0.75 ± 0.27%. This study is the first attempt to analyze the composition of ellagitannins in BRS and to determine optimal extraction conditions for maximum total ellagitannin content from BRS.