A hybrid artificial neural network and multi-objective genetic algorithm approach to optimize extraction conditions of Mentha longifolia and biological activities

Abstract

In this work, artificial neural network coupled with multi-objective genetic algorithm (ANN-NSGA-II) has been used to develop a model and optimize the conditions for the extracting of the Mentha longifolia (L.) L. plant. Input parameters were extraction temperature (40-70 degrees C), extraction time (4-10 h), and extract concentration (0.25-2 mg/mL) while total antioxidant status (TAS) and total oxidant status (TOS) values of extracts were output parameters. The mean absolute percentage error (MAPE) of selected ANN model was determined as 1.434% and 0.464% for TAS and TOS, respectively. The results showed that the optimum extraction conditions were as follows: extraction temperature of 54.260 degrees C, extraction time of 7.854 h, and extract concentration of 0.810 mg/mL. The biological activities and phenolic contents of the extract obtained under determined optimum extract conditions were determined. TAS and TOS values of extract were determined as 6.094 +/- 0.033 mmol/L and 14.050 +/- 0.063 mu mol/L, respectively. Oxidative stress index (OSI) as 0.231 +/- 0.002, total phenolic content (TPC) as 123.05 +/- 1.70 mg/g and total flavonoid content (TFC) as 181.84 +/- 1.97 mg/g. Anti- acetylcholinesterase value and anti-butyrylcholinesterase value of the extract was determined as 42.97 +/- 0.87 and 60.52 +/- 0.80 mu g/mL, respectively. In addition, 11 phenolic compounds, namely acetohydroxamic acid, gallic acid, catechin hydrate, 4-hydroxybenzoic acid, caffeic acid, vanillic acid, syringic acid, 2-hydoxycinamic acid, quercetin, luteolin and kaempferol, were determined. It was observed that the extract of M. longifolia produced under optimum conditions exhibited strong biological activities. These results indicate that ANN coupled NSGA-II was an effective method for the optimization extraction conditions of M. longifolia.