Exercise Benchmark Dose Modelling

MVEN10 Risk Assessment in Environment and Public Health

Author

Zheng Zhou

Published

September 22, 2025

Instructions

This exercise is designed to 1) help you evaluate and improve your understandings of key concepts of benchmark dose modeling, 2) gain practical experience with established benchmark dose modeling tools.

Submission requirements

No report is required for this exercise.

Question 1: NOAEL/LOAEL

You are provided with a dose-response data set from a real study. Download the data.

The study has a control group and three dose groups. For each group, the following data is measured: number of subjects, group average response, group standard deviation, and the p value of the difference in average response compared with the control group. You can find the data by the file “example_data_dose_response.txt”.

Tasks

(1) Explain if this data can derive a NOAEL. If yes, what is it? If no, why?

(2) Explain if this data can derive a LOAEL. If yes, what is it? If no, why?

Question 2: BMD modeling

Using the same dose-response data provided in question. Recall the three steps of BMD modeling. Perform the three steps in the existing BMD tools and generate reports.

Tasks

(1) Use EFSA PROAST for BMD modeling. You can use the web version: https://proastweb.rivm.nl/ or the R-package version: https://www.rivm.nl/en/proast. Generate a report.

(2) Based on the report from (a), find the following information: 1) number of models fitted, 2) AIC of each model, 3) CED, CEDL and CEDU of each model

(3) Use US EPA BMDS for BMD modeling: https://www.epa.gov/bmds/bmds-online. Generate a report.

(4) Based on the report from (c), find the following information: 1) number of models fitted, 2) P-value, AIC and reason of recommendation of each model, 3) BMD, BMDL and BMDU of each model

Question 3: Evaluate dose-response models between NOAEL, LOAEL and the BMDs from different tools

Gather your results of NOAEL/LOAEL (Question 1) and BMDs from different tools (Question 2).

Tasks

Discuss one strengths and one weaknesses of the NOAEL/LOAEL approach and the BMD approach, respectively.

You may consider the following hints:

  1. What information is needed for each approach?

  2. If there are changes in study design, e.g., dose level setup, how will each approach be affected?

  3. How difficult it is to implement each approach?

  4. Does the sample size have impacts on the estimates from the two approaches?

  5. Can a NOAEL be derived from a LOAEL? from a BMD?

  6. What is a benchmark response and how it is chosen?

  7. What are the differences between BMD models?

  8. How are uncertainty in the data considered in the two approaches?