Related Experiment Video
Updated: Aug 5, 2026
Ultrafast Lignin Extraction from Unusual Mediterranean Lignocellulosic Residues
Published on: March 9, 2021
Machine learning-assisted green extraction of polysaccharides from Syringa oblata Lindl leaf residue using natural
Shuang Jiang1, Shulu Zhang2, Jingting Han2
1Department of Pharmaceutical Analysis and Analytical Chemistry of College of Pharmacy of Harbin Medical University. Harbin 150081, China; Department of Pharmacology of College of Pharmacy, Harbin Medical University, Harbin, Heilongjiang 150081, China.
This study integrates machine learning with natural deep eutectic solvents (NADES) to optimize polysaccharide extraction from plant waste. The developed method enhances extraction efficiency and provides a sustainable strategy for agricultural residue valorization.
Area of Science:
- Green Chemistry
- Biotechnology
- Materials Science
Background:
- Underutilized plant resources offer significant potential for sustainable development and agricultural by-product valorization.
- Developing efficient extraction methods is crucial for unlocking the value of these resources and promoting a circular economy.
Purpose of the Study:
- To establish a green and efficient ultrasound-assisted natural deep eutectic solvent (NADES) extraction strategy for polysaccharides from Syringa oblata Lindl (S. oblata) leaf residue.
- To integrate machine learning for predicting and optimizing polysaccharide extraction yield and process parameters.
Main Methods:
- Ultrasound-assisted extraction using NADES.
- Application of multiple machine learning models (including XGBoost) for prediction and optimization.
- SHAP analysis to determine the contribution of process parameters.
- Density Functional Theory (DFT) calculations to elucidate the extraction mechanism.
- Structural characterization and antioxidant activity assays of extracted polysaccharides.
Main Results:
- The XGB machine learning model achieved high prediction accuracy (R² > 0.92) for polysaccharide yield.
- Optimal extraction conditions were determined: 51.4% NADES water content, 42.9 mL/g DSR, 41.7 min ultrasonic time, and 250.0 W ultrasonic power.
- SHAP analysis indicated NADES to S. oblata leaf residue ratio (DSR) as the most influential parameter (58.57%).
- DFT calculations revealed enhanced binding energy and hydrogen-bond interactions between NADES and polysaccharides, promoting extraction.
- Extracted polysaccharides (SOLP) showed significant ABTS and hydroxyl radical scavenging activities.
Conclusions:
- A data-driven framework integrating machine learning, DFT, and NADES extraction was successfully established.
- This approach significantly improves polysaccharide extraction efficiency and prediction accuracy.
- The study presents a promising strategy for the high-value utilization of natural polysaccharides from agricultural residues.
