You are here:
Consensus Modeling in Support of a Semi-Automated Read-Across Application (SOT)
Citation:
Judson, R. Consensus Modeling in Support of a Semi-Automated Read-Across Application (SOT). Presented at Society of Toxicology Annual Meeting, New Orleans, LA, March 13 - 17, 2016. https://doi.org/10.23645/epacomptox.5084230
Impact/Purpose:
Slide or platform presentation at the SOT 2016 annual meeting. Introduce Read-across as a widely used technique to help fill data gasps in risk assessment. With increasing availability of large amounts of computable in vitro and in vivo data on chemicals, it should be possible to build a variety of computer models to help guide a risk assessor in doing a read-across assessment,
Description:
Read-across is a widely used technique to help fill data gaps in a risk assessment. With the increasing availability of large amounts of computable in vitro and in vivo data on chemicals, it should be possible to build a variety of computer models to help guide a risk assessor in doing a read-across assessment. We are developing a suite of “local models” which take available data in all local neighborhoods of chemical space. (The space of chemicals being evaluated covers most of the ~30K for which there is known use.) These are a variety of machine learning and Bayesian models that make probabilistic calls regarding specific endpoints based on availability of data in the local neighborhood. Different models use different descriptor sets, distance metrics and scoring functions. The models are then combined into a consensus, which attempts to minimize or cancel out the bias inherent in each individual model. The performance of the models in each local neighborhood for each endpoint is then evaluated statistically. The models are finally incorporated into a prototype read-across dashboard. This talk will discuss the modeling approach, will walk through case studies and demonstrate the dashboard itself. This abstract does not necessarily reflect U.S. EPA policy
URLs/Downloads:
DOI: Consensus Modeling in Support of a Semi-Automated Read-Across Application (SOT)
JUDSON SOT 2015 LOCAL MODELS.PDF (PDF, NA pp, 1615.238 KB, about PDF)
JUDSON SOT READ ACROSS TALK ABSTRACT.PDF (PDF, NA pp, 15.78 KB, about PDF)