Telecommunications and other digital technology companies that collect user location data can make this data available to health authorities and researchers. Such location data, in anonymised and aggregate format, can be used to map population movements in order to anticipate needs (e.g. identify risk areas) and plan public health resources, as well as to check whether social distancing measures are effective.
Telecommunications and other companies such as Google, Facebook and Uber have long compiled and shared aggregate location data, including for the purpose of fighting pandemics. During the current Covid-19 crisis, a number of companies have started to share aggregate location data and related analytics that show where and when people are gathering. For example, Google has released Community Mobility Reports showing population mobility patterns in 131 countries to help assess the effectiveness of self-isolation rules. According to Google, the reports are created with aggregate, anonymised sets of data from users who have turned on the location history setting on their accounts (which is turned off by default). Facebook has also made maps on population movements available, based on user location and social connections to inform disease forecasting efforts.
Anonymised, aggregate data may be useful for mapping population movements and thus to assess general adherence to confinement measures (which may prompt further measures from law enforcement authorities). However, reducing the spread of the virus may require more precise location data, together with additional information. A study on the 2014 west African Ebola crisis, questioned the effectiveness of location tracking in tackling epidemics, arguing that location data are most useful when cross-referenced with other data (e.g. testing and diagnostics data).
The advantage of using anonymised, aggregate data for mapping population movements is that this approach raises fewer concerns about fundamental rights, such as privacy and data protection. However, the risk of re-identifying the subjects of anonymised data is always present. A growing number of studies show the limitations of existing anonymisation practices, particularly when datasets are mined using machine learning.