Efficient Machine-Learning-Based New Tools to Design Eutectic Mixtures and Predict Their Viscosity DOI

Stella Christodoulou,

Camille Cousseau,

Emmanuelle Limanton

и другие.

ACS Sustainable Chemistry & Engineering, Год журнала: 2024, Номер 12(52), С. 18537 - 18554

Опубликована: Дек. 17, 2024

The development of models that accurately predict the formation eutectic mixtures (EMs, including well-known deep solvents) and their viscosity is imperative to save time in synthesizing new solvents. We developed reliable machine-learning-based classifiers able discern between noneutectic (non-EM) regressors an EM. A experimental data set 219 EMs, 384 non-EMs, 1450 points at different temperatures water contents provided used challenge several models, defined both by algorithm descriptors. top-performing EM/non-EM classifier yields accuracy 92%, best regressor achieves predictions with a mean absolute error 2.2 mPa·s; extrapolation capabilities latter were assessed on additional measurements outside range training set, revealing good low viscosities. SHapley Additive exPlanations (SHAP) was employed as eXplainable Artificial Intelligence (XAI) technique quantify input feature contributions model output. These results represent significant step forward developing robust highly accurate for determining viscosity.

Язык: Английский

A data driven machine learning approach for predicting and optimizing sulfur compound adsorption on metal organic frameworks DOI Creative Commons

Mohsen Shayanmehr,

Sepehr Aarabi,

Ahad Ghaemi

и другие.

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Янв. 24, 2025

This study employed some machine learning (ML) techniques with Python programming to forecast the adsorption capacity of MOF adsorbents for thiophenic compounds namely benzothiophene (BT), dibenzothiophene (DBT), and 4,6-dimethyl (4,6-DMDBT). Five ML models were developed help a dataset containing 676 rows correlate adsorbent features, conditions, adsorbate characteristics sample's sulfur capability. Among approaches, MLP model achieved best performance low mean squared error (MSE) 0.0032 on test set 0.0021 training relative (MRE) 15.26% set. Also, Random Forest yielded higher MSE 0.0045 MRE 17.83%. Feature importance analysis was performed by utilizing shapely additive plan (SHAP) method, findings revealed that "initial concentration sulfur" (SHAP value 0.51) "contact time" 0.37) crucial factors influenced desulfurization process efficiency. Additionally, comparative features network classified into three primary categories: characteristics, characteristics. Consequently, condition identified as most significant group compared others. Finally, optimization indicated maximum DBT 161.6 mg/g Zr-based could be when including BET, TPV, pore size, oil/adsorbent ration, temperature tuned around 756 m2/g, 0.955 cm3/g, 5.96 nm, 449.85 g/g, 20.1 °C, respectively.

Язык: Английский

Процитировано

2

A novel interpretable machine learning and metaheuristic-based protocol to predict and optimize ciprofloxacin antibiotic adsorption with nano-adsorbent DOI
Yunus Ahmed,

Akser Alam Siddiqua Maya,

Parul Akhtar

и другие.

Journal of Environmental Management, Год журнала: 2024, Номер 370, С. 122614 - 122614

Опубликована: Окт. 8, 2024

Язык: Английский

Процитировано

13

Optimizing Photocatalytic Dye Degradation: A Machine Learning and Metaheuristic Approach for Predicting Methylene Blue in Contaminated Water DOI Creative Commons
Yunus Ahmed, Krishna Dutta,

Sharmin Nahar Chowdhury Nepu

и другие.

Results in Engineering, Год журнала: 2024, Номер unknown, С. 103538 - 103538

Опубликована: Дек. 1, 2024

Язык: Английский

Процитировано

12

Solvent Screening for Separation Processes Using Machine Learning and High-Throughput Technologies DOI Creative Commons

Justin P. Edaugal,

Difan Zhang, Dupeng Liu

и другие.

Chem & Bio Engineering, Год журнала: 2025, Номер 2(4), С. 210 - 228

Опубликована: Март 5, 2025

As the chemical industry shifts toward sustainable practices, there is a growing initiative to replace conventional fossil-derived solvents with environmentally friendly alternatives such as ionic liquids (ILs) and deep eutectic (DESs). Artificial intelligence (AI) plays key role in discovery design of novel development green processes. This review explores latest advancements AI-assisted solvent screening specific focus on machine learning (ML) models for physicochemical property prediction separation process design. Additionally, this paper highlights recent progress automated high-throughput (HT) platforms screening. Finally, discusses challenges prospects ML-driven HT strategies optimization. To end, provides insights advance future

Язык: Английский

Процитировано

0

Advanced Ciprofloxacin Quantification: A Machine Learning and Metaheuristic Approach Using Ultrasensitive Chitosan-Gold Nanoparticle Based Electrochemical Sensor DOI
Yunus Ahmed, Tahmina Akter,

Meherunnesa Prima

и другие.

Journal of environmental chemical engineering, Год журнала: 2024, Номер unknown, С. 115094 - 115094

Опубликована: Дек. 1, 2024

Язык: Английский

Процитировано

3

Efficient Machine-Learning-Based New Tools to Design Eutectic Mixtures and Predict Their Viscosity DOI

Stella Christodoulou,

Camille Cousseau,

Emmanuelle Limanton

и другие.

ACS Sustainable Chemistry & Engineering, Год журнала: 2024, Номер 12(52), С. 18537 - 18554

Опубликована: Дек. 17, 2024

The development of models that accurately predict the formation eutectic mixtures (EMs, including well-known deep solvents) and their viscosity is imperative to save time in synthesizing new solvents. We developed reliable machine-learning-based classifiers able discern between noneutectic (non-EM) regressors an EM. A experimental data set 219 EMs, 384 non-EMs, 1450 points at different temperatures water contents provided used challenge several models, defined both by algorithm descriptors. top-performing EM/non-EM classifier yields accuracy 92%, best regressor achieves predictions with a mean absolute error 2.2 mPa·s; extrapolation capabilities latter were assessed on additional measurements outside range training set, revealing good low viscosities. SHapley Additive exPlanations (SHAP) was employed as eXplainable Artificial Intelligence (XAI) technique quantify input feature contributions model output. These results represent significant step forward developing robust highly accurate for determining viscosity.

Язык: Английский

Процитировано

2