Using Machine Learning to Predict Adverse Effects of Metallic Nanomaterials to Various Aquatic Organisms

Abstract

The wide production and use of metallic nanomaterials (MNMs) leads to increased emissions into the aquatic environments and induces high potential risks. Experimentally evaluating the (eco)toxicity of MNMs is time-consuming and expensive due to the multiple environmental factors, the complexity of material properties, and the species diversity. Machine learning (ML) models provide an option to deal with heterogeneous data sets and complex relationships. The present study established an in silico model based on a machine learning properties-environmental conditions-multi species-toxicity prediction model (ML-PEMST) that can be applied to predict the toxicity of different MNMs toward multiple aquatic species. Feature importance and interaction analysis based on the random forest method indicated that exposure duration, illumination, primary size, and hydrodynamic diameter were the main factors affecting the ecotoxicity of MNMs to a variety of aquatic organisms. Illumination was demonstrated to have the most interaction with the other features. Moreover, incorporating additional detailed information on the ecological traits of the test species will allow us to further optimize and improve the predictive performance of the model. This study provides a new approach for ecotoxicity predictions for organisms in the aquatic environment and will help us to further explore exposure pathways and the risk assessment of MNMs.

New Insights on the Formation of Nucleation Mode Particles in a Coastal City Based on a Machine Learning Approach

“…Three metrics were employed to evaluate the performance of a model: a higher coefficient of determination ( R 2 ), a lower root-mean-squared error (RMSE), and a lower mean absolute error (MAE) indicate better model performance (refer to Text S4 of the Supporting Information) . A permutation test, iterated 10 times, assessed cross-validation coefficients ( Q 2 ; refer to Text S5 of the Supporting Information) via linear regression and correlation analysis to detect overfitting. , …”

“…57 A permutation test, iterated 10 times, assessed cross-validation coefficients (Q 2 ; refer to Text S5 of the Supporting Information) via linear regression and correlation analysis to detect overfitting. 59,60 2.3.2. Feature Importance.…”

Nanosafety: a Perspective on Nano‐Bio Interactions

“…However, there are remaining challenges with respect to the environmental and human health impacts of nanomaterials. Hence, while there has been significant progress in assessing the likely exposure of ecological receptors to the most widely employed nanomaterials, and while their effects on aquatic ecosystems have been assessed, albeit with considerable uncertainty, [ 62 ] unexpected risks are likely to come from novel, multi‐component nanomaterials, engineered to make them more mobile and more bioavailable. Even though inadvertent (occupational) exposure, as well as intentional (consumer and medical) exposure to nanomaterials, is on the rise, the types of nanomaterials that are currently released to the environment are rather limited.…”

Application of Machine Learning in Nanotoxicology: A Critical Review and Perspective

“…(1) removal of entire observations with missing data; (2) filling the missing gaps with representative values, such as mean, median, mode and expert knowledge, 20 (3) interpolation using neighbor values; 59 and (4) imputation using ML-based methods. 23,60,61 It is recommended to explore the imputation methods to fill the data gaps, as it has the least error and results in the best prediction accuracy.…”

“…(2) The scope of prediction has expanded from metal and metal oxides NMs to include carbonbased NMs and organic NMs., 17,18 although predictive model development beyond the first generation of NMs is still notoriously challenging; (3) In terms of data sources, databases of significant sizes have been developed, 19 although significantly more effort in strengthening the databases and knowledgebase is needed; (4) the selection of descriptors has advanced from considering only structural parameters and physicochemical properties of NMs to incorporating information on exposure conditions and different species of biota, 20 and to considering whole NM descriptors and descriptors as distributions rather than absolute values. 21 Another important in silico approach risk assessment of NMs is quantitative read-across, a data gap filling technique used on unknown or untested NMs.…”