Nanomaterial Risk Forecasting and Risk Characterization Using Bayesian Networks
Monday, April 11, 2011
Eric Money, Ph.D., Post-Doctoral Associate, Center for the Environmental Implications of NanoTechnology (CEINT), Department of Civil and Environmental Engineering.
Abstract: Traditional ecological risk assessments require a great deal of knowledge about the concentrations of a substance in the environment and any resulting toxicity. This information is typically based on a suite of experimental and field measurements to calculate a deterministic risk quotient that is used to inform decision-makers about the need for further examination of potentially ‘risky’ substances. In the case of nanomaterials in the environment, there is a high degree of uncertainty surrounding many of the traditional aspects needed for a risk assessment. Therefore, a novel approach to understanding the potential risks of nanomaterials in the environment is needed to account for this uncertainty and adapt as our knowledge changes. Bayesian networks contain several features that make them a useful tool for this situation.
This presentation provides an in depth discussion of how Bayesian networks work and the unique features that make them a promising new tool for risk forecasting and risk characterization of nanomaterials. The results of an 18-month structural elicitation are presented in the form of a comprehensive, probabilistic risk forecasting model for nanomaterials we have termed FINE (Forecasting the Impacts of Nanomaterials in the Environment). The case study of nano-Ag in an aquatic environment is illustrated via scenario testing and derived expressions for predicting the exposure potential, dose, and risk of nano-Ag. The presentation concludes with a discussion of ongoing work and the general utility of the FINE model for providing research feedback, in addition to its risk forecasting abilities.