Answers are grounded in this paper and cite their sources.
This study developed an integrated response surface methodology (RSM) and machine learning (ML) framework to model and optimize ultrasonic–microwave-assisted extraction (UMAE) of vanillin from cured Vanilla planifolia beans. A three-factor Box–Behnken experimental design involving microwave power, extraction time, and ethanol concentration was employed to generate data for both RSM and ML models, including Random Forest Regression (RFR), Support Vector Regression (SVR), and Extreme Gradient Boosting (XGBoost). The optimal UMAE conditions were identified as 312.4 W microwave power, 4.2 min extraction time, and 67.1% v/v ethanol. Under these conditions, the process achieved a vanillin yield of 2.95 ± 0.07% and a concentration of 29.1 ± 0.5 mg g −1 , representing a 58% improvement over single-mode microwave or ultrasonic extraction. Enhanced process sustainability was observed, with specific energy consumption reduced to 11.4 MJ kg −1 and apparent diffusivity increased to 5.6 × 10 −9 m 2 s −1 , indicating intensified mass transfer and improved energy efficiency. Among all evaluated models, the RFR algorithm demonstrated the highest predictive accuracy, with an R 2 value of 0.982 and RMSE of 0.014. Feature importance analysis identified microwave power as the most influential variable affecting extraction performance at 48.3%, followed by extraction time at 31.5% and ethanol concentration at 20.2%. Experimental verification conducted near the predicted optimum showed deviations below 3%, confirming good process reproducibility. Overall, the integrated RSM–ML approach enabled consistent optimization with deviations below 2.5% and provides a robust and data-driven framework for process intensification in sustainable food processing.
No takes yet. Share an insight, caveat, or question.
Sujinda et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: