通过微波辐射利用超分子低共熔溶剂从薄荷中可持续提取植物化学物质：基于CatBoost驱动特征分析的深入洞察

Sustainable extraction of phytochemicals from Mentha arvensis using supramolecular eutectic solvent via microwave Irradiation: Unveiling insights with CatBoost-Driven feature analysis.

作者信息

Naseem Zubera, Qadir Muhammad Bilal, Bentalib Abdulaziz, Khaliq Zubair, Zahid Muhammad, Ahmad Fayyaz, Nadeem Nimra, Javaid Anum

机构信息

Department of Textile Engineering, National Textile University, Faisalabad 37610, Pakistan; Department of Chemistry, University of Agriculture Faisalabad 38040, Pakistan.

Department of Textile Engineering, National Textile University, Faisalabad 37610, Pakistan.

出版信息

Ultrason Sonochem. 2025 Apr;115:107300. doi: 10.1016/j.ultsonch.2025.107300. Epub 2025 Mar 4.

The present study revealed the higher extraction potential of sustainable choline chloride (ChCl) and ethylene glycol (EG) based deep eutectic solvent (DES) from Mentha arvensis via microwave irradiation. The categorical boosting (CatBoost) machine learning model was applied to optimize the extraction process against time (4-8 min), microwave power (160-320 W), and biomass quantity (1-2.0 g/10 mL) with DES. The experimentally optimized TPC 124 ± 4.0 mg GAE/g, TFC 79 ± 3.0 mg QE/g, and DPPH radical inhibition 90 ± 4.0 % evaluated in 6 min at 240 W with 1.0 g biomass. The lowest average relative errors of 0.402 % (TPC), 0.863 % (TFC), and 0.597 % (DPPH) for train and 0.679 % (TPC), 0.685 % (TFC) and 0.480 % (DPPH) for test data showed the consistency with the predicted values. The partial dependence and feature importance revealed the contributing impact of parameters for optimizing the extraction. The average contribution percentage of each predictor to the responses revealed that time contributed 32.5 % (TPC), 35.9 % (TFC), and 18.6 % (DPPH); microwave power contributed 26.7 % (TPC), 25.5 % (TFC), and 44.2 % (DPPH); while biomass contributed 40.8 % (TPC), 38.6 % (TFC), and 37.2 % (DPPH). The significant antibacterial (S. aureus = 25.5 ± 1.4 mm and E. coli = 23.5 ± 1.4 mm) with MICs (S. aureus = 50 ± 2.5 µg/mL and E. coli = 100 ± 1.5 µg/mL) and antifungal potential (F. solani = 22.5 ± 1.4 mm, A. niger = 23.5 ± 0.8 mm), with MIC (F. solani = 100 ± 0.4 µg/mL and A. niger = 50 ± 0.5 µg/mL) of optimized extracts recorded by DES. The DES would be the best alternative to traditional organic solvents based on higher extraction efficiency and sustainability.

本研究揭示了通过微波辐射从薄荷中提取可持续的氯化胆碱（ChCl）和乙二醇（EG）基深共熔溶剂（DES）的潜力更高。应用分类提升（CatBoost）机器学习模型，针对时间（4 - 8分钟）、微波功率（160 - 320瓦）和生物质数量（1 - 2.0克/10毫升）与DES来优化提取过程。实验优化后的总酚含量（TPC）为124±4.0毫克没食子酸当量/克，总黄酮含量（TFC）为79±3.0毫克芦丁当量/克，在240瓦、1.0克生物质的条件下6分钟内评估的DPPH自由基抑制率为90±4.0%。训练数据的最低平均相对误差分别为0.402%（TPC）、0.863%（TFC）和0.597%（DPPH），测试数据的最低平均相对误差分别为0.679%（TPC）、0.685%（TFC）和0.480%（DPPH），表明与预测值一致。偏依赖和特征重要性揭示了各参数对优化提取的贡献影响。每个预测变量对响应的平均贡献百分比表明，时间对总酚含量（TPC）的贡献为32.5%、对总黄酮含量（TFC）的贡献为35.9%、对DPPH自由基抑制率的贡献为18.6%；微波功率对总酚含量（TPC）的贡献为26.7%、对总黄酮含量（TFC）的贡献为25.5%、对DPPH自由基抑制率的贡献为44.2%；而生物质对总酚含量（TPC）的贡献为40.8%、对总黄酮含量（TFC）的贡献为38.6%、对DPPH自由基抑制率的贡献为37.2%。优化提取物的显著抗菌性能（金黄色葡萄球菌抑菌圈直径 = 25.5±1.4毫米，大肠杆菌抑菌圈直径 = 23.5±1.4毫米），其最低抑菌浓度（金黄色葡萄球菌 = 50±2.5微克/毫升，大肠杆菌 = 100±1.5微克/毫升）以及抗真菌潜力（茄腐镰孢菌抑菌圈直径 = 22.5±1.4毫米，黑曲霉抑菌圈直径 = 23.5±0.8毫米），最低抑菌浓度（茄腐镰孢菌 = 100±0.4微克/毫升，黑曲霉 = 50±0.5微克/毫升）由DES记录。基于更高的提取效率和可持续性，DES将是传统有机溶剂的最佳替代品。