FUEL
·
2025
·
IF 7.5
Songtao Wu,Modi Guan,Xiaohan Wang,Jing Zhang,Yuhang Zhou,Xiu Huang,Bin Pan
Wettability is a pivotal parameter to determine hydrocarbon reserves and production in shale reservoirs, typically characterized quantitatively by shale-oil-brine contact angles. This parameter is a complex function of shale composition, fluid properties, temperature etc. and is therefore very difficult for conventional theoretical methods to predict accurately and efficiently. Therefore, herein machine learning methods [eXtreme gradient boosting (XGBoost) and the Shapley additive explanations (SHAP)] were integrated to predict this parameter and analyze its sensitivity. To make up the shortage of available data in literature (only 162 data points), another 100 data points about shale/mineral-alkane-water contact angles were measured under in-situ reservoir conditions using the sessile droplet method, thus totally 262 data points were used for machine learning. Experimental results showed that shale, quartz, mica and albite became more hydrophilic with increasing temperature from 25 ℃ to 70 ℃, while K-feldspar and dolomite demonstrated the opposite trend; shale-alkane-water contact angle increased from 60° to 149° (thus wettability shifted from water-wet to oil-wet) with increasing TOC content from 1.9 wt% to 10 wt%. The XGBoost model demonstrated superior predictive accuracy than the gradient boosting regressor and support vector machine models (e.g. R 2 is 0.913, 0.876 and 0.623, respectively). The SHAP sensitivity analysis revealed that brine ionic strength, TOC content, calcite content and quartz content were the four most influential factors affecting wettability. This work presents an efficient artificial intelligence method for shale wettability prediction, which is beneficial for hydrocarbon reserves estimation and production in shale reservoirs.
研究领域:
石油工程
·
地质学
·
人工智能
✓ 已验证
产品关联
DOI
高影响因子
CERAMICS INTERNATIONAL
·
2024
·
IF 5.1
Xiao Zong,Shuangshuang Wu,Kunji Lin,Jintao Zhang,Yehua Li,Donglin Lu,Xin Deng,Shengguo Lu,Jihua Qiu,Yuanzhi Shao,Shanghua Wu
Improving the thermal conductivities and bending strengths of ceramic substrates is crucial for their applicability as key components of integrated chips. Herein, we report the latest study in which the thermal conductivity and bending strength of aluminum nitride (AlN) ceramics were systematically predicted using high-precision prediction model approached with a machine learning (ML) method based on extreme gradient enhancement method (XGBoost). The ML model was used to rank the effects of the process parameters, and SHapley Additive exPlanations (SHAP) was employed to quantify the contributions of different factors. Sintering additives were found to have the predominant influence on the thermal conductivity and bending strength; particularly, those with cationic radii within 0.085–0.105 nm enhance both properties. By using Pr 2 O 3 as the sintering additive and adopting the preparation conditions recommended by the ML model, we prepared AlN ceramics with thermal conductivity as high as 195.63 W m −1 K −1 and bending strength of 371.60 ± 9.31 MPa, thus satisfying the application requirement of high thermal conductivity . The proposed model is also applicable to alumina and silicon nitride ceramics. This study provides a practicable and once-for-all strategy to realize the entire process from the ML prediction design to the preparation of ceramic materials exhibiting superior thermal conductivity and high strength.
研究领域:
材料科学
·
人工智能
✓ 已验证
产品关联
DOI
高影响因子
SEPARATION AND PURIFICATION TECHNOLOGY
·
2024
·
IF 8.1
Bin Liu,Yanan Yu,Zijian Liu,Zhe Cui,Wende Tian
In this study, a quantitative structure–property relationship (QSPR) model has been designed based on machine learning (ML) to offer a new method to accurately predict carbon dioxide (CO 2 ) solubility in aqueous amine solutions. Molecular descriptors are used to denote representative features of the amine molecular structure supplemented with amine solution concentration, CO 2 partial pressure, and temperature as inputs to the model. The coefficient of determination ( R 2 ) of the well-trained ML model reaches 0.971, and the average absolute deviation (AAD) of independent experimental validation is 4.785 %, effectively demonstrating the model’s reliability and generalization performance. Finally, SHapley Additive exPlanations (SHAP) is adopted to reveal the contribution of different features to the model predictions, making the model more transparent and interpretable. Overall work provides a novel, low-cost, efficient method to predict equilibrium CO 2 solubility in aqueous amine solution and offers a new perspective in developing advanced amine for CO 2 capture.
研究领域:
化学工程
·
环境科学
·
数据科学
✓ 已验证
产品关联
DOI
高影响因子
FOOD CHEMISTRY
·
2025
·
IF 9.8
Fujia Dong,Zhaoyang Ma,Ying Xu,Yingjie Feng,Yingkun Shi,Hui Li,Fukang Xing,Guangxian Wang,Zhongxiong Zhang,Weiguo Yi,Songlei Wang
Veterinary drug residues in meat seriously harm human health. Rapid and accurate detection of veterinary drug residues is necessary to minimize contamination. Taking ofloxacin (OFX) residues in mutton as an example, the near-infrared hyperspectral imaging combined with explainable AI was used to evaluate the importance of feature wavelengths in the convolutional neural network-stacked sparse auto-encoder (CNN-SSAE) model for chemical properties. Based on this, the qualitative (residue identification-residue level identification) and quantitative detection of OFX residues in mutton was realized. The results showed that the accuracy of CNN-SSAE in identifying residue and residue level of OFX was 100% and 93.65%, respectively, and the correlation coefficients for validation (R 2 P ) in quantitative detection of OFX residue was 0.8980. In addition, SHapley Additive exPlanation (SHAP) values were used to identify feature wavelengths that contribute the most in the CNN-SSAE model, which effectively explained the quality attribute information that spectral and chemical values may improve the predicted results in the model decision process. The reliability of the CNN-SSAE model was evaluated by statistical validation methods (F-test and T-test). Finally, the visualization diagram of OFX content distribution was established. This study provides a method reference for explainability detection of veterinary drug residues.
研究领域:
食品科学
·
人工智能
·
生物医学
✓ 已验证
产品关联
DOI
PubMed
高影响因子
CHEMICAL ENGINEERING JOURNAL
·
2025
·
IF 13.2
Ju-Xiang Chen,Jia-Wei Song,Nai-Yun Gao,Yu-Qiong Gao,Yi-Xuan Hu
Sodium p-perfluorous nonenoxybenzenesulfonate (OBS), as a novel perfluorinated compound (PFC), has garnered significant attention due to its widespread application. However, its molecular structure is highly stable, making it difficult to degrade. This experiment employs an ANN model and SHAP interpretation method to optimize DBD plasma for highly effective removal of OBS. Under multi-objective optimization of the ANN model (treatment time: 20 min; pH: 4; conductivity: 36 μS/cm; initial concentration: 37.5 mg/L; input power: 40.5 W), the OBS removal rate and energy yield were 91.8 % and 7.659 g/kW•h, respectively. In SHAP analysis, treatment time has the greatest impact on both removal rate and energy yield. Even in the presence of interfering inorganic anions (NO 3 − , HCO 3 − , and Cl − ), DBD plasma could achieve an effective removal rate of over 60 % for OBS. Free radical quenching experiments, density functional theory (DFT), and LC-MS indicate that OBS undergoes under the influence of •OH, •O 2 − , 1 O 2 , and e aq − , involving desulfonation processes, ether cleavage and perfluorinated chain shortening processes. According to ECOASR and T.E.S.T. toxicity analyses, although some toxic byproducts may be generated during the OBS removal process, the environmental risk was low. In complex real-world aquatic environments, the OBS still achieves an effective removal rate exceeding 70 %, demonstrating the broad application prospects of DBD plasma.
研究领域:
环境科学
·
等离子体化学
·
化学工程
·
水处理技术
·
机器学习应用
·
毒理学
✓ 已验证
产品关联
DOI
高影响因子
CHEMICAL ENGINEERING JOURNAL
·
2026
·
IF 12.5
Lina Chen,Jing Hao,Xiulan Weng,Zuliang Chen
Green one-pot synthesis of Ib-FeNPs using Ipomoea batatas leaf extract. • Organic-Fe cooperation supports adsorption, partial oxidation, and fixation. • SHAP analysis of SVR and RF identifies dosage as the dominant predictor. • Fixed-bed treatment removes about 80% of arsenic from real acid mine drainage. Arsenic-contaminated wastewater poses a persistent threat to public health and ecological safety, while conventional remediation materials often involve trade-offs among performance, sustainability, and economic feasibility. Here, biomolecule-coated iron-based nanoparticles (Ib-FeNPs) were synthesized from Ipomoea batatas ( I. batatas ) leaf extract through a green one-pot route for the simultaneous removal of As(III) and As(V). In a mixed‑arsenic system, Ib-FeNPs achieved removal efficiencies of 92.4% for As(III) and 100.0% for As(V) after 360 min at a dosage of 0.3 g·L −1 . These values exceeded those obtained with bare FeNPs (10.3% and 60.1%), the isolated capping layer (21.5% and 92.0%), and their physical mixture (26.5% and 83.8%). Time-resolved liquid chromatography-atomic fluorescence spectrometry detected the partial conversion of initially added As(III) to As(V). Analysis of the As 3d X-ray photoelectron spectra identified both species on the reacted material. A decrease in the surface Fe(III) fraction accompanied by an increase in Fe(II) was consistent with the participation of Fe-containing sites in redox processes. The component and physical-mixture controls showed that assembly of the Fe-containing phase and plant-derived coating improved simultaneous arsenic removal. Material characterization revealed Fe-containing surface environments and organic functional groups. Liquid chromatography-quadrupole time-of-flight mass spectrometry identified 18 major plant-derived biomolecules. Six regression algorithms were compared using a four-factor experimental dataset, and SHapley additive explanations (SHAP) were applied to the selected models to assess the contributions of operating variables to predicted removal efficiencies within the investigated range. Together, these findings support an adsorption-oxidation-fixation pathway at the process level. Ib-FeNPs also showed good reusability over five cycles and achieved 80.0% total arsenic removal in a fixed-bed experiment with acid mine drainage (AMD). Overall, Ib-FeNPs provide a plant-mediated platform for arsenic-contaminated water remediation, while the integrated machine-learning and characterization framework offers a transferable approach to developing bio-nano adsorbents from underutilized biomass. Download: Download high-res image (131KB) Download: Download full-size image
✓ 已验证
DOI
高影响因子
WATER RESEARCH
·
2024
·
IF 11.4
Yangmin Ren,Mingcan Cui,Yongyue Zhou,Shiyu Sun,Fengshi Guo,Junjun Ma,Zhengchang Han,Jooyoung Park,Younggyu Son,Jeehyeong Khim
Permeable reactive barrier (PRB) is an important groundwater treatment technology. However, selecting the optimal reactive material and estimating the width remain critical and challenging problems in PRB design. Machine learning (ML) has advantages in predicting evolution and tracing contaminants in temporal and spatial distribution. In this study, ML was developed to design PRB, and its feasibility was validated through experiments and a case study. ML algorithm showed a good prediction about the Freundlich equilibrium parameter (R 2 0.94 for K F , R 2 0.96 for n). After SHapley Additive exPlanation (SHAP) analysis, redefining the range of the significant impact factors (initial concentration and pH) can further improve the prediction accuracy (R 2 0.99 for K F , R 2 0.99 for n). To mitigate model bias and ensure comprehensiveness, evaluation index with expert opinions was used to determine the optimal material from candidate materials. Meanwhile, the ML algorithm was also applied to predict the width of the mass transport zone in the adsorption column. This procedure showed excellent accuracy with R 2 and root‐mean‐square‐error (RMSE) of 0.98 and 1.2, respectively. Compared with the traditional width design methodology, ML can enhance design efficiency and save experiment time. The novel approach is based on traditional design principles, and the limitations and challenges are highlighted. After further expanding the data set and optimizing the algorithm, the accuracy of ML can make up for the existing limitations and obtain wider applications.
研究领域:
环境工程
·
水文地质学
·
污染控制
·
环境修复技术
·
应用机器学习
✓ 已验证
DOI
高影响因子
Frontiers in Oncology
·
2025
·
IF 3.3
Yu-jiao Hua,Ying Zhang,Rui-Rong Wu,Juan Lv,Yan Zhang,Yan-yan Chen,Yong-juan Ding,Jing-hua Chen
Background Oxaliplatin-induced peripheral neuropathy (OIPN) poses a significant challenge for patients with colorectal tumor, often resulting in treatment interruption or discontinuation and subsequent treatment failure. Herein, a longitudinal untargeted metabolomic study to reveal the metabolomic profiles and biomarkers associated with the progression of OIPN. Methods A prospective cohort of 129 colorectal cancer patients receiving oxaliplatin-based chemotherapy was stratified into four OIPN severity grades (Level 0-3). Serum samples underwent untargeted LC-MS/MS metabolomic analysis, detecting 521 metabolites. Multivariate statistical models and SHAP-guided random forest algorithms were employed to prioritize biomarkers. Machine learning validation included six classifiers assessed via ROC-AUC. Results The cumulative dose of Oxaliplatin chemotherapy plays an important role in OIPN. At the same time, our findings implied that the occurrence of OIPN may be associated with the progression of the disease and the patients’ tumor markers (CEA, CA19-9, CA72-4), as well as immune response and inflammation (ANC, PLT), and metabolic and liver function abnormalities (GGT and UA) (P<0.05).Multivariate statistical analysis combined with SHAP-guided machine learning identified six biomarkers, including thiabendazole, 1-methylxanthine, imidazol-5-yl-pyruvate, 5-hydroxypentanoic acid, spermidine, and 4’-oxolividamine that consistently distinguished OIPN patients (Level 1-3) from non-OIPN controls (Level 0). Machine learning models, validated across six classifiers, demonstrated near-perfect discrimination for early-stage OIPN (AUC nearly 1). However, differentiation between intermediate OIPN grades (Level 1 vs 2, Level 1 vs 3, Level 2 vs 3) yielded lower predictive accuracy (AUC: 0.549–0.843), likely due to cohort size limitations and reliance on subjective sensory-based grading. Pathway enrichment analysis highlighted dysregulation in ABC transporters, central carbon metabolism in cancer, amino acid metabolism, and linoleic acid metabolism, suggesting potential roles in OIPN pathogenesis. Conclusions These findings suggest that the selected biomarkers could serve as a foundation for the prediction and management of OIPN in colorectal cancer patients.
研究领域:
肿瘤学
·
神经毒理学
·
转化医学
·
药物代谢
·
代谢组学
·
精准医学
·
临床药理学
✓ 已验证
产品关联
DOI
PubMed
高影响因子
ACS Sustainable Chemistry & Engineering
·
2025
·
IF 7.3
Yiming Shen,Jiyizhe Zhang,Xiao Liang,Kai Wang
Electrochemical synthesis, with its mild conditions and high atom economy, offers a sustainable alternative for creating high-value products. However, for electrochemical reactions, commonly involving a larger parameter space than traditional chemical reactions, it is challenging to efficiently identify optimal reaction conditions or investigate the underlying principles. In this article, a two-step framework of AI-guided exploration is proposed for investigating the Co(II)-mediated electro-oxidation of p-xylene. A multiobjective diversity-driven Bayesian optimization approach is applied for the first step to search the space of operation, identifying Pareto-optimal solutions that balance maximizing reaction yield and minimizing Faradaic contribution. These solutions are then analyzed by employing SHAP (SHapley Additive Explanations) and additional experiments in the second step to investigate the reaction mechanisms. The optimized conditions achieved higher yields and lower energy consumption compared to reported studies in the literature while avoiding the use of toxic solvents. The framework demonstrates the advantage of AI-guided optimization to accelerate the research of electrochemical reactions by significantly reducing the cost of experiments and reducing reagent consumption as well as waste generation, presenting a new path for exploring sustainable chemical processes.
研究领域:
电化学
·
有机合成
·
催化化学
·
绿色化学
·
计算化学
·
反应工程
·
可持续能源
✓ 已验证
产品关联
DOI
高影响因子
JOURNAL OF THE AMERICAN CERAMIC SOCIETY
·
2025
·Vol. 108
·期 12
·e70209
·
IF 3.8
Zirui Du,Chaokun Song,Kang Guan,Pinggen Rao,Xiaohui Yang,Longteng Bai,Shengbo Shi,Jing Wang,Yongsheng Liu,Qingfeng Zeng
Improving the ablation resistance of carbon fiber-reinforced silicon carbide (C/SiC) composites is essential to meet the stringent demands of ultra-high-temperature applications, but traditional empirical design approaches are resource-intensive and time-consuming. This study presents a novel data-driven methodology integrating machine learning (ML) with experimental validation to optimize C/SiC composite ablation resistance. Through systematic data preprocessing and feature analysis of 102 experimental samples, an XGBoost regression model was developed, achieving satisfactory prediction accuracy (mean absolute error < ∼0.075, mean squared error < ∼0.015, and coefficient of determination [R 2 ] > ∼0.75) for ablation rate. SHapley Additive exPlanations (SHAP) analysis revealed that modifier properties, particularly standard enthalpy and melting point, predominantly influence ablation resistance. The ML-guided design strategy, implemented through Bayesian optimization, led to the successful fabrication of ZrB 2 -modified C/SiC composites with exceptional ablation resistance. The optimized composite, containing approximately 20.0 vol.% ZrB 2 , achieved a linear ablation rate of 2.083 µm/s under oxyacetylene torch testing, representing a significant improvement over conventional compositions. Microstructural analysis confirmed the formation of a dense SiO 2 ‒B 2 O 3 protective layer, validating the predicted mechanism of enhanced ablation resistance. This work establishes a robust framework for accelerated development of ultra-high-temperature ceramics and demonstrates the efficacy of ML-driven approaches in materials design optimization. An object-oriented software with interactive graphical user interface has been developed. These methodologies have been integrated into an interactive software, Modified C/SiC Ablation Rate intelligent design Software (MARS), creating an efficient tool for the accelerated design of C/SiC composites with tailored ablation performance.
研究领域:
材料科学
·
复合材料工程
·
高温结构材料
·
计算材料学
·
先进陶瓷
✓ 已验证
产品关联
DOI
高影响因子
Advanced Photonics
·
2025
·
IF 18.8
Haoran Chen,Ruike Zhao,Xinyuan Bi,Nan Shen,Xi Mo,Yue Tao,Zhou Chen,Jian Ye
Surface-enhanced Raman spectroscopy (SERS), as a great potential label-free tool in metabolite detection, offers a strategy for rapid bacterial identification. However, it still lacks experimentally supported spectral interpretation at the metabolite level for complex biosamples. We present a SERS-based method for reliable bacterial intracellular metabolic profiling using plasmonic colloids with high rapidness and cost-efficiency. A convolutional neural network model was constructed to accurately classify eight types of bacteria with an overall accuracy as high as 90.44% and identify the key spectral features for classification by Shapley Additive Explanations. Molecule-level interpretation of the SERS metabolic profiles has been further realized in combination with laser desorption/ionization mass spectrometry, evidencing the primary metabolite contribution to the bacterial spectral signatures and molecule-level distinctions among different bacterial types. We provide insights into the mechanism of bacterial identification by label-free SERS and pave the way for interpretable SERS diagnostic tools for various diseases.
研究领域:
分析化学
·
微生物学
·
生物光子学
·
代谢组学
·
机器学习
·
光谱学
✓ 已验证
产品关联
DOI
高影响因子
JOURNAL OF HAZARDOUS MATERIALS
·
2024
·
IF 12.2
Xuyi Wei,Xiaohui Mao,Junwei Han,Wenqing Qin,Hongbo Zeng
Efficient recovery of rare earth elements (REEs) from wastewater is crucial for advancing resource utilization and environmental protection. Herein, a novel nitrogen-rich hydrogel adsorbent (PEI-ALG@KLN) was synthesized by modifying coated kaolinite-alginate composite hydrogels with polyethylenimine through polyelectrolyte interactions and Schiff’s base reaction. Various characterizations revealed that the high selective adsorption capacity of Ho (155 mg/g) and Nd (125 mg/g) on PEI-ALG@KLN is due to a combination of REEs (Lewis acids) via coordination interactions with nitrogen-containing functional groups (Lewis bases) and electrostatic interactions; its adsorption capacity remains more than 85 % after five adsorption-desorption cycles. In waste NdFeB magnet hydrometallurgical wastewater, the recovery rate of PEI-ALG@KLN for Nd and Dy can reach more than 93 %, whereas that of Fe is only 5.04 %. Machine learning prediction was used to evaluate adsorbent properties via different predictive models, with the random forest (RF) model showing superior predictive accuracy. The order of significance for adsorption capacity was pH > time > initial concentration > electronegativity > ion radius, as indicated by the RF model feature importance analysis and SHapley Additive exPlanations values. These results confirm that PEI-ALG@KLN has considerable potential in the selective extraction of REEs from wastewater.
研究领域:
环境科学
·
材料科学
·
化学工程
·
废物管理
·
机器学习
✓ 已验证
产品关联
DOI
PubMed
高影响因子
危险品化学品经营许可证（带存储）