List of Benchmark Dose Models
Most of the models in the following tables were developed by US EPA and are available in BMDS Online (including BMDS Desktop and pybmds), unless otherwise noted.
Some models are flagged as available only in BMDS 2.7. The BMDS 2.7 and BMDS 3.3 software packages are available for download but are no longer supported or updated by EPA.
The BMDS User Guide thoroughly describes these models, including the differences between maximum-likelihood estimation (MLE) and Bayesian modeling.
Maximum Likelihood Estimation (MLE) Models
Model | BMDS 2.7 | BMDS 3.3 | BMDS Online |
---|---|---|---|
Exponential | Yes | Yes | Yes |
Hill | Yes | Yes | Yes |
Linear | Yes | Yes | Yes |
Polynomial | Yes | Yes | Yes |
Power | Yes | Yes | Yes |
Model | BMDS 2.7 | BMDS 3.3 | BMDS Online |
---|---|---|---|
Gamma | Yes | Yes | Yes |
Logistic | Yes | Yes | Yes |
Log-Logistic | Yes | Yes | Yes |
Log-Probit | Yes | Yes | Yes |
Multistage | Yes | Yes | Yes |
Probit | Yes | Yes | Yes |
Weibull | Yes | Yes | Yes |
Quantal Linear | Yes | Yes | Yes |
Dichotomous Hill | Yes | Yes | Yes |
Quantal Models with Background Dose Parameter |
Yes | No | No |
Model | BMDS 2.7 | BMDS 3.3 | BMDS Online |
---|---|---|---|
Nested Logistic | Yes | Yes | Yes |
National Center for Toxicological Research (NCTR) | Yes | No | No |
Rai and Van Ryzin (No longer supported) | Yes | No | No |
Specialized Models
Model | BMDS 2.7 | BMDS 3.3 | BMDS Online |
---|---|---|---|
Bayesian Model Averaging | No | Yes | Yes |
Multi-tumor (MS_Combo) | Yes | Yes | Yes |
Models No Longer Supported
- Repeated Response Measures / ToxicoDiffusion (BMDS 2.7 only)
- Documentation (Science Inventory Database)
- ten Berge Concentration x Time (BMDS 2.7 only; superseded by CatReg 3.0)
- Documentation (Science Inventory Database)
- Multistage Weibull (MSW) Time-to-Tumor
- Software, documentation, and peer review comments (Science Inventory Database). The model executable is a DOS-based application.
- Also available is an R-based graphical tool to generate diagnostic plots with MSW outputs and assist users to assess goodness-of-fit for models fitted using MSW.