Full Content is available to subscribers

Subscribe/Learn More  >

Selecting Nanoparticle Properties to Mitigate Risks to Workers and the Public: A Machine Learning Modeling Framework to Compare Pulmonary Toxicity Risks of Nanomaterials

[+] Author Affiliations
Jeremy M. Gernand, Elizabeth A. Casman

Carnegie Mellon University, Pittsburgh, PA

Paper No. IMECE2013-62687, pp. V015T12A016; 20 pages
  • ASME 2013 International Mechanical Engineering Congress and Exposition
  • Volume 15: Safety, Reliability and Risk; Virtual Podium (Posters)
  • San Diego, California, USA, November 15–21, 2013
  • Conference Sponsors: ASME
  • ISBN: 978-0-7918-5644-4
  • Copyright © 2013 by ASME


Due to their size and unique chemical properties, nanomaterials have the potential to interact with living organisms in novel ways, leading to a spectrum of negative consequences. Though a relatively new materials science, already nanomaterial variants in the process of becoming too numerous to be screened for toxicity individually by traditional and expensive animal testing. As with conventional pollutants, the resulting backlog of untested new materials means that interim industry and regulatory risk management measures may be mismatched to the actual risk. The ability to minimize toxicity risk from a nanomaterial during the product or system design phase would simplify the risk assessment process and contribute to increased worker and consumer safety.

Some attempts to address this problem have been made, primarily analyzing data from in vitro experiments, which are of limited predictive value for the effects on whole organisms. The existing data on the toxicity of inhaled nanomaterials in animal models is sparse in comparison to the number of potential factors that may contribute to or aggravate nanomaterial toxicity, limiting the power of conventional statistical analysis to detect property/toxicity relationships. This situation is exacerbated by the fact that exhaustive chemical and physical characterization of all nanomaterial attributes in these studies is rare, due to resource or equipment constraints and dissimilar investigator priorities.

This paper presents risk assessment models developed through a meta-analysis of in vivo nanomaterial rodent-inhalational toxicity studies. We apply machine learning techniques including regression trees and the related ensemble method, random forests in order to determine the relative contribution of different physical and chemical attributes on observed toxicity. These methods permit the use of data records with missing information without substituting presumed values and can reveal complex data relationships even in nonlinear contexts or conditional situations.

Based on this analysis, we present a predictive risk model for the severity of inhaled nanomaterial toxicity based on a given set of nanomaterial attributes. This model reveals the anticipated change in the expected toxic response to choices of nanomaterial design (such as physical dimensions or chemical makeup). This methodology is intended to aid nanomaterial designers in identifying nanomaterial attributes that contribute to toxicity, giving them the opportunity to substitute safer variants while continuing to meet functional objectives.

Findings from this analysis indicate that carbon nanotube (CNT) impurities explain at most 30% of the variance pulmonary toxicity as measured by polymorphonuclear neutrophils (PMN) count. Titanium dioxide nanoparticle size and aggregation affected the observed toxic response by less than ±10%. Difference in observed effects for a group of metal oxide nanoparticle associated with differences in Gibbs Free Energy on lactate dehydrogenase (LDH) concentrations amount to only 4% to the total variance. Other chemical descriptors of metal oxides were unimportant.

Copyright © 2013 by ASME



Interactive Graphics


Country-Specific Mortality and Growth Failure in Infancy and Yound Children and Association With Material Stature

Use interactive graphics and maps to view and sort country-specific infant and early dhildhood mortality and growth failure data and their association with maternal

Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In