Abstract:Predictive models for micropollutant removal by membrane separation are highly desirable for the design and selection of appropriate membranes. While machine learning (ML) models have been applied for such purposes, their reliability might be compromised by data leakage due to inappropriate data splitting. More importantly, whether ML models can truly understand the mechanisms of membrane separation has not been revealed. In this study, we evaluate the capability of the XGBoost model to predict micropollutant removal efficiencies of reverse osmosis and nanofiltration membranes. Our results demonstrate that data leakage leads to falsely high prediction accuracy. By utilizing a model interpretation method based on the cooperative game theory, we test the knowledge of XGBoost on the mechanisms of membrane separation via quantifying the contributions of input variables to the model predictions. We reveal that XGBoost possesses an adequate understanding of size exclusion, but its knowledge of electrostatic interactions and adsorption is limited. Our findings suggest that future work should focus more on avoiding data leakage and evaluating the mechanistic knowledge of ML models. In addition, high-quality data from more diverse experimental conditions, as well as more informative variables, are needed to improve the accuracy of ML models for predicting membrane performance.The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.1c04041.Details on PCA and K-fold cross-validation; input variables; absolute errors of the model predictions with and without data leakage (with detailed data); absolute errors of the model predictions as a function of input variables; comparison of MAE of our study with the literature; predictions of the model with different split ratios (i.e., training/validation data: testing data are 90/10, 80/20, or 70/30); and 10 replicates of the model predictions with data split ratios of 90/10 (PDF)Micropollutant removal efficiencies of commercial and self-fabricated membranes (XLSX)Micropollutant removal efficiencies of commercial and self-fabricated membranes (XLSX)This article has not yet been cited by other publications.

Machine Learning Models to Predict Early Breakthrough of Recalcitrant Organic Micropollutants in Granular Activated Carbon Adsorbers

An Experimental Method For Predicting The Adsorption Of Trace Organic Contaminants In Partially Saturated Granular Activated Carbon

Assessing the potential of machine learning methods to study the removal of pharmaceuticals from wastewater using biochar or activated carbon

Machine learning prediction of dye adsorption by hydrochar: Parameter optimization and experimental validation

Insights into the adsorption of pharmaceuticals and personal care products (PPCPs) on biochar and activated carbon with the aid of machine learning

Predicting Aqueous Adsorption of Organic Compounds Onto Biochars, Carbon Nanotubes, Granular Activated Carbons, and Resins with Machine Learning

How machine learning boosts the understanding of organic pollutant adsorption on carbonaceous materials: A comprehensive review with statistical insights

Coupling machine learning and theoretical models to compare key properties of biochar in adsorption kinetics rate and maximum adsorption capacity for emerging contaminants

Machine learning based-model to predict catalytic performance on removal of hazardous nitrophenols and azo dyes pollutants from wastewater

Predicting Micropollutant Removal by Reverse Osmosis and Nanofiltration Membranes: Is Machine Learning Viable?

Applied Machine Learning for Prediction of CO 2 Adsorption on Biomass Waste-Derived Porous Carbons

Machine Learning-Based Prediction of the Adsorption Characteristics of Biochar from Waste Wood by Chemical Activation

Machine learning assisted adsorption performance evaluation of biochar on heavy metal

Application of Machine Learning in Prediction of Pb2+ Adsorption of Biochar Prepared by Tube Furnace and Fluidized Bed

Machine Learning Application for Predicting Key Properties of Activated Carbon Produced from Lignocellulosic Biomass Waste with Chemical Activation

Gaussian Process Regression and Machine Learning Methods for Carbon-Based Material Adsorption

Predicting Cd(II) adsorption capacity of biochar materials using typical machine learning models for effective remediation of aquatic environments

The application of machine learning methods for prediction of heavy metal by activated carbons, biochars, and carbon nanotubes

Machine learning for the adsorptive removal of ciprofloxacin using sugarcane bagasse as a low-cost biosorbent: comparison of analytic, mechanistic, and neural network modeling

Machine learning-driven prediction of phosphorus adsorption capacity of biochar: Insights for adsorbent design and process optimization

Applied Machine Learning For Prediction Of Co2 Adsorption On Biomass Waste-Derived Porous Carbons