A new method of longitudinal diary assembly for human exposure modeling

Area of Science:

• Environmental Health Sciences
• Exposure Science
• Computational Epidemiology

Background:

• Accurate human exposure modeling relies on longitudinal time-activity diaries, but most data are cross-sectional (1 day/person).
• Existing data limitations hinder capturing day-to-day variability crucial for exposure and health risk assessments.
• A method is needed to synthesize longitudinal sequences from cross-sectional activity data.

Purpose of the Study:

• To develop a novel procedure for generating longitudinal time-activity sequences from cross-sectional data.
• To enable better estimation of human pollutant exposure by accounting for intra- and interindividual variability.
• To create adaptable sequences for various exposure modeling applications.

Main Methods:

• A new method ranks 1-day activity diaries based on a user-selected key variable.
• The method stochastically assembles longitudinal diaries targeting specific statistics (D and A) for variance and autocorrelation.
• The D statistic measures within- and between-person variance; the A statistic measures day-to-day autocorrelation.

Main Results:

• The method closely attains target D and A statistics for exposure analysis periods over 30 days.
• Reasonably good attainment of targets was observed for shorter simulation periods.
• Longitudinal diary data suggest D and A statistics are stable over time and potentially across cohorts.

Conclusions:

• The developed method effectively generates longitudinal time-activity sequences from cross-sectional data.
• This approach enhances the accuracy of human exposure modeling and subsequent health risk assessments.
• The method is adaptable to various cohort definitions and diary pool assignments in exposure models.