Optimisation of heat reflux extraction of gallic acid, quercetin and chlorogenic acid from Stevia rebaudiana (Bertoni) leaves by using response surface methodology and adaptive neuro-fuzzy inference system modelling

Stevia rebaudiana Bertoni, belongs to the Asteraceae family is a perennial shrub and widely recognised as a natural sweetener. This plant contains steviol glycosides and antioxidant-rich phytoconstituents. However, optimisation of extraction parameters of antioxidants from this plant are not fully explored yet. The present study aimed to optimise heat-reflux extraction (HRE) using Adaptive Neuro-Fuzzy Inference System (ANFIS) and Response Surface Methodology (RSM). The extraction efficiency was evaluated in terms of antioxidant activity and the yield of gallic acid, quercetin, and chlorogenic acid. Plackett–Burman design (PBD) modelling was used significant parameters followed by Box–Behnken Design (BBD). Quantification of gallic acid, quercetin, and chlorogenic acid was performed using High-Performance Thin-Layer Chromatography (HPTLC) with optimised mobile phases—hexane:ethyl acetate:acetone (16.4:3.6:0.2 v/v), ethyl acetate:glacial acetic acid:formic acid:water (20:2.2:2.2:5.2 v/v), and toluene:ethyl acetate:formic acid (13.5:9:0.6 v/v), respectively. Optimal extraction conditions were established and ANFIS predictions exhibited strong concordance with experimental and RSM outcomes.