Skip to main content

As researchers seek insights into the effects of COVID-19 and non-pharmaceutical interventions (NPIs) on human behavior, mobility data provides an empirical view into how populations move, work, shop, and gather. Through countless analyses, mobility data has revealed fluctuating patterns in travel, commuting, and foot traffic to businesses and other public points of interest throughout the pandemic’s multiple phases and corresponding lockdowns across the world.

Such analyses aid public health and policy officials to make data-driven decisions in setting NPIs and economic reopening policies. However, a key question that has dogged decision makers throughout the pandemic is whether closing businesses and public spaces actually reduces social mixing, or if it simply relocates such behaviors to other venues, such as residential areas. This question also arose within a University of Leeds research group, led by Professor Ed Manley, where they “looked to narrow in on the question of continued mixing between households.”

While mobility data can shed light on these issues in order to improve policy making, a critical question emerges: how can we derive these important insights without sacrificing the privacy of the users who entrust others with their data?

Historically, it was common practice for geospatial data providers to directly share de-identified and privacy-enhanced data with researchers, who would then hold copies of the data and perform analyses locally. To protect sensitive locations, such as residential areas, the data must be shared under strict licensing agreements, where the data itself is pre-processed to add privacy-preserving noise.  As a result, any analysis on changing rates of social mixing in residential neighborhoods is impossible.

By completely flipping the traditional model of data sharing on its head, the Spectus platform is not only enabling these important analyses, but is strengthening privacy protections and governance frameworks in the process. Rather than sharing granular data directly with researchers, Spectus’ PaaS (platform as a service) solution provides researchers with a secure and auditable ‘sandbox’ environment. In this way, users can query Spectus’ first-party data, while receiving highly aggregated outputs in return.

For example, the analysis conducted by the University of Leeds research team queried the number of times fully de-identified users visited locations labeled as residential areas that weren’t their own, but at a city-level spatial resolution. In this way, the team was able to capture information on the rates of change of cross-household visits across UK cities, without ever accessing household-level data that could explain who came from which households, which specific addresses they visited, or for what purposes.

For Manley and his research group, it is not about seeing individual behaviors. Instead, their aim is “to better understand where and when the policies fail, so we can improve future public guidance and support with greater specificity.”

This article was originally authored while the Spectus Data Clean Room product was known as Cuebiq Workbench.