Optimization of extraction and bioactivity detection of celery leaf flavonoids using BP neural network combined with genetic algorithm and response

Abstract

In the present study, ultrasound-assisted extraction was employed to extract the general flavone from celery leaves using response surface methodology and BP neural network model with a genetic algorithm (GA). The effects of temperature, time, solid-liquid ratio, and ethanol concentration on the extraction results were assessed by Box-Behnken design. Further optimization of the process was performed by GA-BP. Our results showed that the optimal conditions were an ethanol concentration of 70.31%, a temperature of 67.2 °C and an extraction time of 26.6 min. In addition, significant antioxidant activity and in vitro bacteriostasis were observed. We found that the total flavonoids of the celery leaves exerted a strong inhibitory effect on <i>Escherichia coli</i>, <i>Staphylococcus aureus</i>, and <i>Bacillus subtilis</i>. Additionally, considerable DPPH· and ·OH scavenging effects were exerted by flavonoids. Therefore, flavonoids from celery leaves can be considered natural antioxidants and bacterial inhibitors.