Using anonymous location data from mobile devices, researchers were able to forecast COVID-19 outbreaks in Connecticut towns.
According to the researchers, the unique approach used in the study might assist health authorities better manage COVID-19 outbreaks in communities and allocate testing resources.
Data scientists and epidemiologists from the Yale School of Public Health, the Connecticut Department of Public Health, the US Centers for Disease Control and Prevention, and Whitespace Ltd., a spatial data analytics organization, collaborated on the project, according to Medical Xpress.
The accuracy with which researchers were able to detect episodes of high frequency near personal contact (defined as a radius of 6 feet) down to the municipal level in Connecticut was the key to the results. To minimize probable COVID-19 transmission, the CDC recommends keeping a distance of at least six feet between persons.
“The primary route for transmission of SARS-CoV-2, the virus that causes COVID-19, is close contact between people,” said study lead author Forrest Crawford, an associate professor of biostatistics at Yale School of Public Health as well as an associate professor of ecology and evolutionary biology, management, statistics, and data science at Yale.
“We used mobile device geolocation data to quantify close interpersonal interaction within a 6-foot radius anywhere in Connecticut over the course of a year,” Crawford said. “This project provided epidemiologists and policymakers in Connecticut with knowledge into people’s social distancing behavior throughout the state.”
Other research have employed “mobility measurements” as proxies for social distancing behavior and COVID-19 transmission. However, such analysis might be faulty.
Crawford said, “Mobility measurements often assess distance traveled or time spent away from a place, such as your house.” “We all know, however, that you may travel about a lot and yet not become close to other people. As a result, mobility measurements aren’t a good indicator of transmission risk. Close contact, we believe, better predicts infections and local epidemics.”
The conclusions are based on an examination of geolocation data from Connecticut mobile devices from February 2020 to January 2021. No personally identifying information was gathered, and all of the data was anonymized and aggregated.
Based on geolocation data, an unique algorithm calculated the likelihood of near contact events (times when mobile devices were within six feet of each other) throughout the state. This data was then fed into a conventional COVID-19 transmission model, which was used to forecast COVID-19 case numbers not just throughout Connecticut, but also in particular towns, census tracts, and census block groupings.
The researchers claimed to have correctly anticipated an initial wave of COVID-19 infections in Connecticut from March to April 2020, a decline in statewide incidence from June to August, and isolated outbreaks in select Connecticut municipalities in August and September.
To monitor the spread of COVID-19, many health authorities now depend on general surveillance data such as the number of verified cases, hospitalizations, and fatalities. However, this process might take days or weeks longer than real disease spread. According to the study, analyzing close personal contact rates is substantially quicker.
“We developed a contact rate in this study that can reveal high-contact conditions that are likely to spawn local outbreaks and areas where residents are at high transmission risk days or weeks before the resulting cases are detected through testing, traditional case investigations, and contact tracing,” Crawford said.
