Comparison of Machine Learning Approaches with a General Linear Model To Predict Personal Exposure to Benzene

Investigating the impact of high-altitude on vehicle carbon emissions: A comprehensive on-road driving study

This study addresses the literature gap concerning accurately identifying vehicle carbon emission characteristics in high-altitude areas. Utilizing a portable emission measurement system (PEMS) for real-world testing, we quantified the influence of altitude on carbon emissions from light-duty gasoline (LDGV) and diesel vehicles (LDDV). The Random Forest (RF) algorithm was employed to analyze the complex nonlinear relationships between altitude, meteorological conditions, driving patterns, and carbon dioxide (CO2) emissions, enabling predictions across different altitudes. The results showed that CO2 emissions progressively increase with elevation. Furthermore, as altitude increases, combustion efficiency declines, and the overall impact of driving conditions on emission rates diminishes. Altitude and meteorological factors significantly contributed to CO2 emissions, whereas driving conditions and road grades contributed less. Compared with the COPERT model, the RF model demonstrates strong accuracy in predicting carbon emissions at different altitudes. Specifically, the CO2 emission rate nearly triples as altitude increases from 2.0 km to 4.5 km. This research bridges a critical gap in the understanding carbon emissions from high-altitude vehicles, offering insights into policy development for emission reduction strategies in such regions. Future studies should integrate diverse testing methodologies and comprehensive surveys to validate and extend the findings.

Photocatalytic degradation of drugs and dyes using a maching learning approach

The waste management industry uses an increasing number of mathematical prediction models to accurately forecast the behavior of organic pollutants during catalytic degradation. With the increasing quantity of waste generated, these models are critical for reinforcing the efficiency of wastewater treatment strategies. The application of machine-learning techniques in recent years has notably improved predictive models for waste management, which are essential for mitigating the impact of toxic commercial waste on global water supply. Organic contaminants, dyes, pesticides, surfactants, petroleum by-products, and prescription drugs pose risks to human health. Because traditional techniques face challenges in ensuring water quality, modern strategies are vital. Machine learning has emerged as a valuable tool for predicting the photocatalytic degradation of medicinal drugs and dyes, providing a promising avenue for addressing urgent demands in removing organic pollutants from wastewater. This research investigates the synergistic application of photocatalysis and machine learning for pollutant degradation, showcasing a sustainable solution with promising effects on environmental remediation and computational efficiency. This study contributes to green chemistry by providing a clever framework for addressing present-day water pollution challenges and achieving era-driven answers.

Meta-analysis of engineered nanoparticles dynamic aggregation in freshwater-like systems using machine learning techniques

Predictive algorithms for exposure characterization of engineered nanoparticles (ENPs) in the ecosystems are essential to improve the development of robust nano-safety frameworks. Here, machine learning (ML) techniques were utilised for data mining and prediction of the dynamic aggregation transformation process in aqueous environments using case studies of nZnO and nTiO₂. Supervised ML models using input variables of natural organic matter, ionic strength, size, and ENPs concentration showed poor prediction performance based on statistical metric values of root mean square error (RMSE), mean absolute error (MAE), coefficient of determination (R²), and Nash-Sutcliffe efficiency (NSE) for both types of ENP. On the contrary, algorithms developed using model input parameters of zeta potential, pH, and time had good generalisation and high prediction accuracy. Among the five developed ML algorithms, random forest regression, support vector regression, and artificial neural network generated good prediction accuracy for both data sets. Therefore, the use of ML can be valuable in the development of robust nano-safety frameworks to optimise societal benefits, and for proactive long-term ecological protection.

Mapping the personal PM2.5 exposure of China's population using random forest

Ambient monitoring may cause estimation errors, and wearable monitoring is expensive and labor-intensive when assessing PM_2.5 personal exposure. Estimation errors have limited the development of exposure science and environmental epidemiology. Thus, we developed a scenario-based exposure (SBE) model that covered 8 outdoor exposure scenarios and 1 indoor scenario with corresponding time-activity patterns in Baoding City. The linear regression analysis of the SBE yielded an R² value of 0.913 with satisfactory accuracy and reliability. To apply the SBE model to large-scale studies, we predicted time-activity patterns with the random forest model and atmosphere-to-scenario ratios with the linear regression model to obtain the essential parameters of the SBE model; their R² was 0.65-0.93. The developed model would economize the study expenditure of field sampling for personal PM_2.5 and deepen the understanding of the influences of indoor and outdoor factors on personal PM_2.5. Using this method, we found that the personal PM_2.5 exposure of Chinese residents was 10.50-347.02 μg/m³ in 2020, higher than the atmospheric PM_2.5 concentration. Residents in North and Central China, especially the Beijing-Tianjin-Hebei region and the Fen-Wei Plains, had higher personal PM_2.5 exposure than those in other areas.

Improved Machine Learning Models by Data Processing for Predicting Life-Cycle Environmental Impacts of Chemicals

Machine learning (ML) provides an efficient manner for rapid prediction of the life-cycle environmental impacts of chemicals, but challenges remain due to low prediction accuracy and poor interpretability of the models. To address these issues, we focused on data processing by using a mutual information-permutation importance (MI-PI) feature selection method to filter out irrelevant molecular descriptors from the input data, which improved the model interpretability by preserving the physicochemical meanings of original molecular descriptors without generation of new variables. We also applied a weighted Euclidean distance method to mine the data most relevant to the predicted targets by quantifying the contribution of each feature, thereby the prediction accuracy was improved. On the basis of above data processing, we developed artificial neural network (ANN) models for predicting the life-cycle environmental impacts of chemicals with R² values of 0.81, 0.81, 0.84, 0.75, 0.73, and 0.86 for global warming, human health, metal depletion, freshwater ecotoxicity, particulate matter formation, and terrestrial acidification, respectively. The ML models were interpreted using the Shapley additive explanation method by quantifying the contribution of each input molecular descriptor to environmental impact categories. This work suggests that the combination of feature selection by MI-PI and source data selection based on weighted Euclidean distance has a promising potential to improve the accuracy and interpretability of the models for predicting the life-cycle environmental impacts of chemicals.

Multiple machine learning algorithms assisted QSPR models for aqueous solubility: Comprehensive assessment with CRITIC-TOPSIS

As an essential environmental property, the aqueous solubility quantifies the hydrophobicity of a compound. It could be further utilized to evaluate the ecological risk and toxicity of organic pollutants. Concerned about the proliferation of organic contaminants in water and the associated technical burden, researchers have developed QSPR models to predict aqueous solubility. However, there are no standard procedures or best practices on how to comprehensively evaluate models. Hence, the CRITIC-TOPSIS comprehensive assessment method was first-ever proposed according to a variety of statistical parameters in the environmental model research field. 39 models based on 13 ML algorithms (belonged to 4 tribes) and 3 descriptor screening methods, were developed to calculate aqueous solubility values (log K_ws) for organic chemicals reliably and verify the effectiveness of the comprehensive assessment method. The evaluations were carried out for exhibiting better predictive accuracy and external competitiveness of the MLR-1, XGB-1, DNN-1, and kNN-1 models in contrast to other prediction models in each tribe. Further, XGB model based on SRM (XGB-1, C = 0.599) was selected as an optimal pathway for prediction of aqueous solubility. We hope that the proposed comprehensive evaluation approach could act as a promising tool for selecting the optimum environmental property prediction methods.

Database examination, multivariate analysis, and machine learning: Predictions of vapor intrusion attenuation factors

Traditional soil vapor intrusion (VI) models usually rely on preset conceptual scenarios, simplifying the influences of limiting environmental covariates in determining indoor attenuation factors relative to subsurface sources. This study proposed a technical framework and applied it to predict VI attenuation factors based on site-specific parameters recorded in the United States Environmental Protection Agency (USEPA)'s and the California Environmental Protection Agency (CalEPA)'s VI databases, which can overcome the limitations of traditional VI models. We examined the databases with multivariate analysis of variance to identify effective covariates, which were then employed to develop VI models with three machine learning algorithms. The results of multivariate analysis show that the effective covariates include soil texture, source depth, foundation type, lateral separation, surface cover, and land use. Based on these covariates, the predicted attenuation factors by these new models are generally within one order of magnitude of the observations recorded in the databases. Then the developed models were employed to generate the generic indoor attenuation factors to subsurface vapor (i.e., the 95th percentile of selected dataset), the values of which are different between the USEPA's and CalEPA's databases by one order of magnitude, although comparable to recommendations by the USEPA and literature, respectively. Such a difference may reflect the significant regional disparity in factors such as building structures or operational conditions (e.g., indoor air exchange rates), which necessitates generating generic VI attenuation factors on a state-specific basis. This study provides an alternative for VI risk screens on a site-specific basis, especially in states with a good collection of datasets. Although the proposed technical framework is used for the VI databases, it can be equally applied to other environmental science problems.

Characterization of vapor intrusion sites with a deep learning-based data assimilation method

Appropriate characterization of site soils is essential for accurate risk assessment of soil vapor intrusion (VI). In this study, we develop a data assimilation method based on deep learning (i.e., ES_(DL)) to estimate the distribution of soil properties with limited measurements. Two hypothetical VI scenarios are employed to demonstrate site characterization using the ES_(DL) method, followed by validation with a laboratory sandbox experiment and then one practical site application. The results show that the ES_(DL) method can provide reasonable estimates of the effective diffusion coefficient distributions and corresponding emission rates (into the building) in all four cases. The spatial heterogeneity of site soils can be characterized by considerably enough measurements (i.e., 15 sampling points in the first hypothetical case); otherwise, layered characterization is recommended at the cost of neglecting horizontal heterogeneity of site soils. This new method provides an alternative to characterize VI sites with relatively fewer sampling efforts.

Role of modelling in monitoring soil and food during different stages of a nuclear emergency

In case of a nuclear accident, adequate protection of the public and the environment requires timely assessment of the short- and long-term radiological exposure. Measurements of the radiation dose and the radioactive contamination in the environment are essential for the optimization of radiation protection and the decision making process. In the early phase, however, such measurements are rarely available or sufficient.To compensate for the lack of monitoring data during nuclear emergencies, especially in the early phase of the emergency, mathematical models are frequently used to assess the temporal and spatial distribution of radioactive contamination. During the transition and recovery phase, models are typically used to optimise remediation strategies by assessing the cost-effectiveness of different countermeasures. A prerequisite of course is that these models are fit for purpose. Different models may be needed during different phases of the accident. In this paper, we discuss the role of radioecological models during a nuclear emergency, and give an outlook on the scientific challenges which need to be addressed to further improve our predictions of human and wildlife exposure.