Response Surface and Particle Swarm-Artificial Neural Network Model Optimize the Microwave-assisted Extraction of Paeoniflorin

Abstract

The effects of the following independent variables-micro intensity, extraction time, ethanol concentration, and solvent-to-solid ratio on the extraction yield of paeoniflorin from Paeoniae Radix Rubra were examined by single factor test and Box-Behnken design. The antioxidant activity in vitro of extractions was also assessed. Then, the paeoniflorin extraction process was optimized using the response surface methodology (RSM) and particle swarm optimization-artificial neural network (PSO-ANN). Results showed that the prediction and optimization performance of PSO-ANN was better than RSM, that with the correlation coefficient R2 was 0.9925 and 0.9099, respectively. The optimized extraction conditions by PSO-ANN were as follows: Ethanol concentration (81% v/v), solvent-to-solid ratio (30 mL/g), extraction time (22 s), extractions (5 times), and micro intensity (420 W). Under the optimized parameters, the extraction yield of paeoniflorin was 378.977±1.982 mg PE/g d.w.. The scavenging rates of paeoniflorin extract (100 μg/mL) on DPPH and ABTS+ free radicals were 87.61% and 80.74% respectively, that closed to the positive control. The extract also had a certain reduction ability. The results of this study provide a new method for optimizing the extraction process, as well as provide a theoretical basis for the application of effective components of Paeoniae Radix Rubra as additives.

Keywords

• paeoniflorin
• microwave-assisted extraction
• response surface
• particle swarm optimization-artificial neural network
• antioxidant activity in vitro