Engineered nanomaterials (ENMs) have found their applications in various technologies and consumer products. Manipulation of chemicals at the nanoscale introduces unique characteristics to these materials and makes them desirable for technological applications.
Study: Exploring the potential of in silico machine learning tools for the prediction of acute nanotoxicity of Daphnia magna. Image credit: Michael Traitov/Shutterstock.com
With the increase in ENM production, there have been adverse effects on the environment. Furthermore, it is unfeasible to estimate the risks caused by ENMs every time through in vivo or in vitro experiments. To this end, in silico methods can come to the rescue to perform these assessments.
In a paper published in the journal Chemosphere, we investigated the performance of different machine learning algorithms for predicting a well-defined in vivo toxicity endpoint and identified the important features involved with the in vivo nanotoxicity of Daphnia magna.
The results revealed comparable performances of all algorithms, and predictive performance exceeded approximately 0.7 for all metrics evaluated. In addition, artificial neural network, random forest, and k-nearest neighbor models showed slightly better performance compared to the other algorithm models.
Variable importance analysis performed to understand the importance of input variables revealed that physicochemical properties and molecular descriptors were important in most of the models. On the other hand, the properties related to the exposure conditions gave conflicting results. Thus, machine learning models helped generate in vivo endpoints, even with smaller datasets, demonstrating their reliability and robustness.
Role of machine learning in nanotechnology
Nanotechnology has emerged as a key technology with implications in agriculture, medicine and food industries. Thus, ENMs are more attractive than their larger counterparts due to their outstanding features due to their smaller size.
Despite their advantages, ENMs have also caused effects on the environment, affecting environmental health and safety, calling for an environmental risk assessment associated with ENMs. However, such evaluation by in vivo or in vitro testing for all fabricated nanoforms is impractical.
The challenge in risk assessment is not only due to the extensive production and applications of ENM, but also to the great diversity of materials. To this end, chemical modification at the nanometric scale can modulate the physicochemical properties and the consequent toxicity profile of the materials.
Recent advances in machine learning provided new tools to extract new knowledge from large data sets and to acquire small data sets more efficiently. Nanotechnology researchers use machine learning tools to tackle challenges in many fields. Because of its support for complex interactions, machine learning can help predict the toxicological effects of ENMs using large datasets.
The field of nanotoxicology lacks standardized procedures for representing common ontologies for measuring ENM properties. However, models from limited data sets can help generate the key nanotoxicological descriptors. Nanotoxicological models based on machine learning developed so far focused on endpoints such as viability or cytotoxicity.
In Silico Machine Learning Tools for Daphnia Magna Nanotoxicity Prediction
Despite considerable efforts, several obstacles to in silico modeling of nanotoxicological effects still exist due to limited data availability and poor data curation. Therefore, better agreement on data quality, experimental protocols and availability is vital to acquire homogeneous data across studies.
In the present work, we investigated the performance of machine learning algorithms to predict the in vivo nanotoxicity of metallic ENMs towards Daphnia magna. Several models were generated from the sources obtained from the fixed assets data, which were congruent with the Organization for Economic Co-operation and Development (OECD) principles. Additionally, limitations were overcome to obtain consistent data for modeling by applying different data curation methods.
Among the six machine learning models generated from OECD, neural network, random forest, and k-nearest neighbor algorithms showed the highest performance, while the other models showed relatively similar performance. This indicates that machine learning is more suitable for in silico modeling of in vivo nanotoxicity than the actual algorithm. In addition, the key descriptors modulating the toxicity of metallic ENMs towards Daphnia magna were also studied based on the generated machine learning models.
Conclusion
In summary, machine learning algorithms were performed to predict the in vivo nanotoxicity of metallic ENMs. Daphnia magna toxicity data collected for metal ENMs were analyzed using six classification machine learning models based on OECD principles.
The results revealed that artificial neural networks, random forests, and k-nearest neighbor algorithms had the highest performances, which were in line with previous reports in the literature. On the other hand, the relative differences in other algorithm models were comparatively small. These results demonstrated the suitability of machine learning for in silico modeling of in vivo nanotoxicity.
Furthermore, feature importance analysis using machine learning algorithms revealed conflicting results across all models, with physicochemical properties and molecular descriptors as significant features within the models. The results showed that models with small datasets with few physicochemical properties and molecular descriptors lead to machine learning models with good predictive performance.
reference
Balraadjsing, S., Peijnenburg, W JGM, Vijver, MG (2022) Exploring the potential of in silico machine learning tools for the prediction of acute nanotoxicity of Daphnia magna. Chemosphere https://www.sciencedirect.com/science/article/pii/S0045653522024237
Disclaimer: The views expressed here are those of the author expressed privately and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and Conditions of Use for this website.