Optimization and analysis of ultrasound-assisted solvent extraction of bioactive compounds from Boerhavia diffusa Linn. using RSM, ANFIS and machine learning algorithm

Boerhavia diffusa L., a flowering plant is traditionally used as a therapy for protecting against several diseases. The present study aims to identify the optimal extraction conditions for ultrasound-assisted solvent extraction (USAE) for retrieving highest amount of biologically important compounds from B. diffusa L. through RSM (Response Surface Methodology) and validated by ANFIS (adaptive neuro-fuzzy inference system), and MLA (machine learning algorithm) models. Numerous extraction parameters have played a major role in the extraction process for obtaining maximum yield of bioactive compounds. The four majorly contributing independent parameters, ethanol concentration (X₁: 60–70%), temperature (X₂: 35–45 °C), particle size (X₃: 300–500 µm), and ultrasonic-exposure time (X₄: 15–20 min), and at five levels (− 2, − 1, 0, + 1, + 2) concerning dependent parameters, TPC (y₁), TFC (y₂), %DPPH*sc (y₃), %ABTS*sc (y₄) and %H₂O₂*sc (y₅)) were chosen. The optimal extraction condition from RSM was observed at X₁ = 67.5–70%, X₂ = 35 °C, X₃ = 300 µm and X₄ = 20 min; under this situation, y₁ = 312.59–316.271 mg gallic acid equivalents (GAE)/g, y₂ = 138.748–142.052 mg rutin equivalents (RU)/g and their antioxidant potentials (y₃ = 59.98–61.621%, y₄ = 76.762–78.642%, and y₅ = 64.623–62.362%) have been noted. ANFIS and MLA were used to authenticated the optimized extraction parameters of RSM. Many experimental values were well-matched with the predicted values of ANFIS and MLA. A well-fitted quadratic model was obtained. Further, GC–MS analysis performed to identify the compounds present in the optimized extract yielded 17 compounds. In silico molecular docking study was done to predict the nephroprotective effect of the compounds present in the optimized extract.