Friday, April 2, 2021

Identifying and preventing environmental causes of disease requires estimating long-term personal exposures. Most environmental exposures do not have valid biomarkers and studies therefore rely on external exposure assessment methods.  Approaches are split broadly into methods for modeling exposures for large populations versus measuring exposures for small populations. New technologies and resulting big data offer tremendous opportunity for unifying these approaches and improving long-term personal exposure prediction at scales needed for population-based research.  In this presentation I will provide examples from ongoing research projects, such as: (1) leveraging existing individual time-activity data from Google Location History to estimate long-term environment exposures; (2) integrating image-based deep learning models to assess environmental exposures; and (3) applying new air pollution sensors to large epidemiological studies. I will make the case that a multi-disciplinary approach is needed to combine these types of technologies to improve personal exposure measures, enhance epidemiological research, and identify new prevention opportunities.