Optimization of ultrasonic-assisted extraction of total flavonoids from Oxalis corniculata by a hybrid response surface methodology-artificial neural network-genetic algorithm (RSM-ANN-GA) approach, coupled with an assessment of antioxidant activities

Abstract

The objective of this research endeavor is to refine the ultrasonic-assisted extraction technique for total flavonoids from Oxalis corniculata (TFO), utilizing a synergistic approach combining response surface methodology (RSM) and artificial neural network integrated with genetic algorithm (RSM-ANN-GA). The optimized extraction parameters determined through RSM yielded a TFO concentration of 13.538 mg g⁻¹ under the following conditions: an ethanol concentration of 61.95%, a liquid–solid ratio of 41.06 mL g⁻¹, an ultrasonic power setting of 351.57 W, and an ultrasonic exposure duration of 58.95 minutes. Conversely, the RSM-ANN-GA approach identified an even more refined set of conditions, achieving a TFO concentration of 13.7844 mg g⁻¹, with an ethanol concentration of 58.93%, a liquid–solid ratio of 41.16 mL g⁻¹, an ultrasonic power of 350.22 W, and an ultrasonic exposure time of 58.18 minutes. These findings underscore the superior predictive accuracy and enhanced extraction efficiency offered by the RSM-ANN-GA model over the conventional RSM method. Furthermore, the study demonstrated that TFO possesses a potent antioxidant effect, as evidenced by its ability to scavenge DPPH, hydroxyl, and superoxide anion free radicals in vitro, highlighting its potential as a valuable source of natural antioxidants.