You are here:
A Consumer's Guide To Benchmark Dose Models (BMDS): Results Of U.S. EPA Testing Of 14 Dichotomous, 8 Continuous, And 6 Developmental Models (Presentation)
Citation:
Gift, J., J. Caldwell, AND R. Setzer. A Consumer's Guide To Benchmark Dose Models (BMDS): Results Of U.S. EPA Testing Of 14 Dichotomous, 8 Continuous, And 6 Developmental Models (Presentation). Presented at Annual SRA Meeting, December 1999.
Description:
Benchmark dose risk assessment software (BMDS) was designed by EPA to generate dose-response curves and facilitate the analysis, interpretation and synthesis of toxicological data. Partial results of QA/QC testing of the EPA benchmark dose software (BMDS) are presented. BMDS provides a Windows 95/98 interface to run seven (7) models (Gamma, Logistic/Log=Logistic, Multistage, Probit/Log-Probit, Quantal-Linear, Quantal-Quadratic, and Weibull) that are appropriate for the analysis of dichotomous data, four (4) models (Linear/Polynomial, Power, Hill and Hybrid) for continuous dose response data and three (3) nested developmental toxicology data (NLogistic, NCTR, and Rai & Van Ryzin). One part of EPA QA/QC testing of the BMDS performed during the summer of 1999 involved comparing the BMDS models to other, correlate models developed outside of EPA. Each of the EPA and non-EPA models were applied to 100 dichotomous or continuous dose-response data sets appropriate for evaluation by the model. Data sets were obtained primarily from cancer and non-cancer summaries in the EPA's integrated Risk Information System (IRIS) database. Textual and graphical results of the model runs (e.g., maximum likelihood values for the fitted model and p-values, shape of the fitted curve) are presented and evaluated for the purpose of determining the strengths and weaknesses of each model.