Response surface methodology and artificial neural network modelling for optimization of solid-liquid extraction and rapid HPTLC analysis of glycyrrhizin in Glycyrrhiza glabra root

Glycyrrhizin, a major bioactive compound of Glycyrrhiza glabra has been used in traditional medicine system to treat several diseases. In the past few decades, its demand has been increased significantly for the active ingredient glycyrrhizin as well as for being a zero calorie sweetener. In the current study, extraction of glycyrrhizin from dry roots of G. glabra was established by reflux method. Quantitative estimation of glycyrrhizin content in the extracted samples was analysed by using high performance thin layer chromatography (HPTLC) system. In this study, response surface methodology (RSM) and artificial neural network (ANN) were used for the first time as a model to optimize the extraction conditions of glycyrrhizin from G. glabra roots in order to compare and establish effective prediction models. The extraction process variables including the time, temperature, solvent composition, particle size, solvent to solute ratio, pH and extraction steps were optimized using a Plackett-Burman design (PBD), Central Composite design (CCD), artificial neural network (ANN) and Multiple Layer Perceptron (MLP) on glycyrrhizin yield. The optimum extraction conditions were 55°C (extraction temperature), 45 min (extraction time), 55% (v/v) (ethanol concentration) and 30 ml/g (solvent to solute ratio) to obtain the maximum glycyrrhizin yield. The contributions of the quadratic model with high determination coefficient (R 2 =0.9806) were observed and showed good consistency between the experimental value 7.6 mg/g (0.76%) and predicted value 7.51 mg/g (0.751%). Response values were close to the predicted values with prediction accuracy as ANN > RSM, indicating that ANN model has higher prediction accuracy than RSM. The study suggests that RSM and ANN model system can be manipulated for the optimization and production of valuable bioactive compounds.