A sensitivity analysis of a human exposure model using the Sobol method

Area of Science:

• Environmental Health
• Exposure Science
• Computational Toxicology

Background:

• The Air Pollutants Exposure Model (APEX) is a validated tool used by the U.S. Environmental Protection Agency (EPA) for assessing population-level inhalation exposure to airborne pollutants.
• For over 30 years, APEX and its predecessor (pNEM) have been instrumental in human exposure assessments.
• Understanding model sensitivities is crucial for prioritizing data collection and ensuring the reliability of simulation results.

Purpose of the Study:

• To apply variance decomposition using the Sobol method to identify key inputs and processes driving variability in APEX exposure and dose estimates.
• To elucidate the most influential factors affecting simulated population exposure and dose for children aged 5-18 years.
• To inform the EPA and model users on prioritizing resources for data collection and analysis.

Main Methods:

• Application of the Sobol method, a variance decomposition technique, for sensitivity analysis of the APEX model.
• Examination of daily-average ozone exposure and inhaled ozone dose for children aged 5-18 years.
• Quantification of the influence of various input parameters, including activity diaries, microenvironmental factors, home location, age, sex, and body mass.

Main Results:

• Activity diaries and microenvironmental parameters (air-exchange rate, decay rate) were identified as the most influential factors affecting ozone exposure and dose.
• Aggregate main-effect indices (MEIs) for these parameters were 0.818 for daily-average ozone exposure and 0.469 for daily-average inhaled ozone dose.
• Home location, age, sex, and body mass showed more modest influences on exposure and dose, with aggregate MEIs of 0.079 and 0.307, respectively.

Conclusions:

• The sensitivity analysis highlights the critical role of activity patterns and microenvironmental conditions in determining children's ozone exposure and dose.
• These findings support the EPA's efforts to refine exposure assessments by focusing on the most impactful variables.
• The sensitivity analysis also served as a quality assurance measure, aligning model sensitivities with known physical properties.