Practices and Trends of Machine Learning Application in Nanotoxicology.

Irini Furxhi,Martin Mullins,Finbarr Murphy,Athanasios Arvanitis,Craig A Poland

doi:10.3390/nano10010116

Abstract

Machine Learning (ML) techniques have been applied in the field of nanotoxicology with very encouraging results. Adverse effects of nanoforms are affected by multiple features described by theoretical descriptors, nano-specific measured properties, and experimental conditions. ML has been proven very helpful in this field in order to gain an insight into features effecting toxicity, predicting possible adverse effects as part of proactive risk analysis, and informing safe design. At this juncture, it is important to document and categorize the work that has been carried out. This study investigates and bookmarks ML methodologies used to predict nano (eco)-toxicological outcomes in nanotoxicology during the last decade. It provides a review of the sequenced steps involved in implementing an ML model, from data pre-processing, to model implementation, model validation, and applicability domain. The review gathers and presents the step-wise information on techniques and procedures of existing models that can be used readily to assemble new nanotoxicological in silico studies and accelerates the regulation of in silico tools in nanotoxicology. ML applications in nanotoxicology comprise an active and diverse collection of ongoing efforts, although it is still in their early steps toward a scientific accord, subsequent guidelines, and regulation adoption. This study is an important bookend to a decade of ML applications to nanotoxicology and serves as a useful guide to further in silico applications.

Highlights

Nanomaterials/nanoforms (NMs) display high heterogeneity regarding their physicochemical (p-chem) properties, quantum-mechanical properties, and, as such, their toxicological impact, which renders assessing their risk a case-by-case challenge
This paper provides an extensive up-to-date review focusing on the techniques that are used to predict a human health and/or environmental outcome including selection of algorithms and the employed performance metrics and applicability domain methods
In order to investigate Machine Learning (ML) models in the field of nanotoxicology, we explored several sources of the peer-reviewed scientific literature and reports executing a systematic Boolean search with key terms, such as “nanoparticle,” “nanomaterial,” “in silico,” “computational,” “machine learning,” “model,” and “nanotoxicity.” These were used to form defined multiple search strings, which were applied to publicly available electronic search engines (Google Scholar, ScienceDirect, Web of Science, and PubMed) with the aim of being able to discover studies that implement an ML model to predict nanotoxicity (Table 1)

Summary

Introduction

Nanomaterials/nanoforms (NMs) display high heterogeneity regarding their physicochemical (p-chem) properties, quantum-mechanical properties, and, as such, their toxicological impact, which renders assessing their risk a case-by-case challenge. Traditional hazard assessment relies mostly on in vivo testing that poses technical challenges, e.g., regarding the validity of extrapolation to humans, ethical dilemmas, and comes with high resource demands in cost and time [1]. Such an approach is not conducive to efficient identification and mitigation of possible risks, especially within emerging technologies where the pace of development is rapid. Developing the knowledge base needed for robust modelling for predicting NM properties, exposure, and hazard potential would improve the design of new materials while maximizing utility and minimizing adverse biological effects (safe-by-design) [2,3]. In silico tools are not yet accepted by regulators as a stand-alone solution due to a lack of standardization, but as a complementary tool [6,7]

Methods

Results

Discussion

Conclusion