Practices and Trends of Machine Learning Application in Nanotoxicology

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.

Predicting In Vitro Neurotoxicity Induced by Nanoparticles Using Machine Learning

“…For example, surface area, hydrodynamic size and zeta potential measured in different media were absent in almost half our data samples. Nevertheless, RF has good performance even with missing values [37], highlighting the RF selection [19] and demonstrating the strength of ML tools to bypass missing knowledge. Researchers still encounter data shortness and lack of harmonized protocols, and theoretical understanding further complicates making data reproducible.…”

“…As measurement units and magnitude range differ and can affect the optimization of the model during training, data was normalized [31]. Data normalization was conducted on the numeric inputs to enhance model performance [19]. We used different normalization techniques such as log10, z-score and min-max for each input individually while assessing the skewness.…”

“…Bootstrapping in general refers to sampling of data with replacement maintaining the same data distribution [38,39]. In our cases, we implemented bootstrapping in two steps; first an oversampling was applied to the balanced dataset to increase dataset size; then, we randomly sampled the dataset 10 times to different sets of 90% (training dataset for internal validation) and 10% (external validation) of the data [19]. Oversampling was applied through a supervised filter that doubled the number of instances.…”

“…Diverse ML tools have been developed during the last two decades that explore the prediction of toxicological properties or the adverse effects of NPs [18]. Quantitative structure-activity/toxicity relationships (QSARs/ QSPRs) are among the most widely used practices [19]. ML does not require deterministic insights; bypassing in depth comprehension of the interactions within a system, it constructs a computational predictor bridging input data directly to the outcome.…”

Engineered Nanomaterials: The Challenges and Opportunities for Nanomedicines

“…Thus, the study of machine learning application in toxicology studies remains a dynamic and diverse manner, where preliminary standardized methodology requires a substantial period of research duration to subside. 127 …”