Statistical properties of longitudinal time-activity data for use in human exposure modeling

Area of Science:

• Environmental Health
• Exposure Science
• Human Activity Patterns

Background:

• Characterizing human exposure to pollutants requires understanding longitudinal time-activity patterns.
• Accurate human activity algorithms are essential for EPA's exposure modeling efforts.

Purpose of the Study:

• To present longitudinal, multi-season time-activity diary study results for eight working adults.
• To improve the parameterization of human activity algorithms in exposure modeling.
• To analyze diversity (D) and lag-one autocorrelation (A) statistics for various locations.

Main Methods:

• Conducted a four-season longitudinal time-activity diary study with eight working adults.
• Collected data on time spent in outdoor, motor vehicle, residential, and other-indoor locations.
• Calculated diversity (D) and lag-one autocorrelation (A) statistics, analyzing day-type, season, temperature, and gender differences.

Main Results:

• Overall D and ICC values ranged from 0.08-0.26.
• Mean population rank A values ranged from 0.19-0.36.
• Intra-individual variability exceeded inter-individual variability, with low day-to-day correlations among locations.

Conclusions:

• Human activity patterns exhibit significant intra-individual variability, challenging traditional exposure modeling.
• Current exposure models often underestimate population exposure distributions and health risks by not accounting for these behavioral characteristics.
• Findings highlight the need for improved algorithms that incorporate individual variability in time-activity for more accurate exposure and health risk assessments.