Skip to main content

Mahsa Mirzaei

Supervisors:

Dr Finbarr Murphy Dr Martin Mullins

Working Title of Thesis:

Nano-Toxicology Risk Assessment and Machine Learning

Abstract

Nanotechnology is a field of science and engineering that deals with structures having at least one of their three dimensions less than 100 nanometers (nm). The size range of nanomaterials (NMs) matches proteins or even small viruses. The use of NMs has raised safety concerns, as their small size eases accumulation in and interaction with biological tissues, such as in discoveries of drugs, chemicals, and materials. The interactions between the NMs might explain the toxicological properties of the combination in humans and environments.

As a result, a standard assay or framework for nanotoxicity in-vivo and in-vitro is needed to ensure the standardized assessment of nano-safety. In-vivo assessment is difficult due to high costs, low efficiency, ethical dilemmas, and the validity of extrapolating between species and long-term effects of NMs are problematic to experimentally reproduce in cells or in-vivo. It also is unrealistic to fulfill all safety assessments by in-vivo testing considering the tremendous number of NMs; the need to minimize costs and reduce animal suffering all through toxicity assessment of NMs has encouraged the use of alternative or complementary methods such as in-silico for toxicological evaluation. Many predicting models that are employed to assess NMs toxicity have been reported. Subsequently, we require the knowledge for robust modeling to allow materials and designs with maximized utility and minimized toxicity (safe-by-design). In the era of data, machine learning (ML) algorithms have played a crucial role by improving the predictions of toxicological effects and the design of engineered nanomaterials (ENMs) and strategies to eliminate exposure and minimize risks. As the size of the information increases this artificial intelligence (AI) methods are of great value to analyze this massive amount of data. The exceptional capacities of computers can be exploited to fulfill approaches impractical with in-vivo and in-vitro methods. ML applications are transforming our capacity to predict toxicities and informing safe design.

It is both challenging and stimulating as a research issue, to find suitable in-silico models capable of extracting and discovering knowledge from NMs data sources. The main concerns for nano-informatics are collection, maturity, standards, and accessibility of data, in other words, lack of harmonized data on nanotechnology. And there is a need to ensure consistent representation of nanotechnological data. This research will fill the gaps concerning data quality and knowledge integration. Moreover, it will offer promising insights into the parameters and variables that mostly affect the toxicity of nanoparticles. Finally, it will give an advantage to building a new computational tool to enable the prediction of the NMs hazard and exposure dimensions. This research will be of great value to the scientific community and stakeholders by offering tools developed under standardized guidelines.