Demographic characterization of human dynamics models with time-use and travel surveys

Understanding human dynamics – how people live, move, and interact – is critical to support public service delivery in fields including transportation, energy, and health. Digital trace data provides rich information about occupancy at points of interest (POIs) and flows among locations at varying times. However, notions of "who" moves and interacts, at which times, and in which places, are difficult to obtain using current data sources. To address this issue, we present a novel approach for characterizing human dynamics by incorporating demographic factors including life course, living arrangement, and socioeconomic status. Including such information in human dynamics models enables more direct impact assessments of spatial planning and policy interventions – for example, promoting situational awareness of epidemics by estimating variation in social contact networks at different times of day. To preserve individual privacy, this research is built upon synthetic populations and activity trajectories derived from public-use data sources. We develop a spatial microsimulation model that 1) generates synthetic populations from American Community Survey data, 2) assigns agents from the synthetic population activity profiles derived from the American Time Use Survey and National Household Travel Survey, and 3) simulates trip chains (e.g., home, school, errands) for agents based on multimodal travel on real-world transportation networks derived from OpenStreetMap. We present a proof of concept that simulates routine weekday and weekend activities for various demographic segments in United States metropolitan areas, then validate it against observed activities from commercial mobility datasets