Adaptive data-driven age and patch mixing in contact networks with recurrent mobility
What did we do?
We analyzed global positioning system (GPS) mobility data to estimate how often people traveled between Ontario neighborhoods and relative changes to mobility versus pre-pandemic levels. We then combined these data with numbers of daily contacts per person (e.g. household, work, school, & other, stratified by 5-year age groups) estimated by another study, to estimate the average numbers of daily contacts between all combinations of neighborhoods and age groups across Ontario. These contact patterns (or “mixing” patterns) are now being used as inputs for COVID-19 transmission modeling that can help assess the impacts of hotspot-based vaccine prioritization.
Our work, led by Jesse Knight, builds upon work by Arenas et al. (2021) and Prem et al. (2021) to integrate detailed contact patterns by both neighborhoods and age groups. In our approach, people from neighborhoods A and B can contact each other if they meet in neighborhood C, increasing the possibility of transmission between seemingly disconnected neighborhoods. Additionally, we ensure that changes to the daily numbers of different types of contacts (e.g. fewer school contacts in response to school closures) are accurately reflected in overall mixing patterns.