You are here:
Dose-Response Modeling – Benchmark Dose Modeling Approaches Using EPA’s BMDS Online and NIEHS’s ToxicR
Citation:
Davis, Allen, M. Wheeler, J. Zhao, A. Shapiro, T. Blessinger, AND Jeff Gift. Dose-Response Modeling – Benchmark Dose Modeling Approaches Using EPA’s BMDS Online and NIEHS’s ToxicR. Society for Risk Analysis Annual Meeting (Virtual), Tampa, FL, December 04 - 08, 2022.
Impact/Purpose:
Provide training on the use of BMDS Online and ToxicR in dose-response analyses of animal toxicological data.
Description:
For years, EPA’s Benchmark Dose Software has been available as a stand-alone Windows desktop application for the dose-response analysis of toxicological data for risk assessment. In addition to the updated BMDS Excel release in 3.3; the EPA and NIEHS have expanded dose-response capabilities to Online (Web) and an R-package (free software for statistical computing) environments. BMDS-Online reimplements the existing BMDS software in a web-based application, allowing users to run BMDS on any computer with access to the internet. ToxicR is a R-based Bayesian modeling platform developed by NIEHS/NTP that “untethers” BMDS and other models from standard parameterizations, expanding its capabilities for research applications. This workshop will cover dose-response analyses (frequentist and Bayesian); participants will learn and practice (through hands-on exercises) dose-response modeling of dichotomous and continuous response data using BMDS-Online. Following these introductory analyses, participants will learn and practice the use of Bayesian models, including the application of a Bayesian framework for model averaging using ToxicR. Participants will explore model averaging approaches for dichotomous and continuous data, including new model averaging capabilities for continuous data that include the European Food Safety Authority’s (EFSA) suite of continuous models currently only available in ToxicR. The research functionality and modeling capacity of the ToxicR platform will be demonstrated. Hands-on exercises in ToxicR will allow trainees to be among the first to experience Bayesian modeling practices that may represent the future direction of BMDS. Participants will be shown how to modify prior assumptions and perform sensitivity analyses to investigate the default prior’s effect on a given analysis. Additional features of the package that allow for scripted batch processing, advanced graphics, and custom BMD analysis will also be highlighted. Participants need to bring their own laptops, with the following software installed: · BMDS 3.3 (http://www.epa.gov/bmds); (Windows is required for this portion of the training) · A recent internet browser (Google Chrome is preferred) · R (https://cran.r-project.org) · R-Studio (https://rstudio.com/) · The latest Toxic R package (https://cran.r-project.org/web/packages/ToxicR/index.html) Disclaimer: the views expressed in the abstract are those of the authors and do not necessarily reflect the views or policies of the U.S. EPA or the National Institute of Environmental Health Sciences.