Mobile phone data discourages reopening bars and restaurants
Tracking U.S. mobility points to highest risk areas, Stanford Study finds
BarcelonaThe answer that everyone is waiting for: When will bars and restaurants be able to reopen? What about the cultural sector? What about nightlife? When will it finally be possible to return to a certain normality? Epidemiologists have their numbers: a sustained Rt of approximately 0.7, an ICU occupancy of no more than 300 beds, and an incidence of no more than a thousand cases a day. The problem is to determine what measures need to be taken to achieve this. Research led by the universities of Stanford and Chicago - in addition to the Microsoft research centre - which uses data from 98 million users, identifies the main points of dissemination of covid-19 and the population segments which are most affected. Bars, restaurants and hotels are the establishments with the highest losses, especially in the lower income urban areas. The results have been published in the journal Nature.
The study combines statistical models of coronavirus dissemination and mobility data collected from the mobile phones of 98 million users. The data, conveniently anonymized, as highlighted in the publication, was collected between March 1 and May 1, the period in which the pandemic was most virulent during the first wave. Thanks to this cross-checking of information, researchers have developed "hypotheses" as to where the virus is most transmitted and how racial and economic differences influence its transmission. As expected, racial minorities and lower incomes are the segments of the population who are at risk the most. This occurs because, as described by Sarina Chang, a researcher at Stanford University and lead author of the study, it is in these population groups that it is most difficult to implement containment measures. "They need to work," she exclaims.
What was not so expected was that bars, restaurants, shops, malls, gyms or places of worship such as churches, points that could intuitively be considered as risky, would play such a big part, at least during the first wave. All of these establishments play a "disproportionately important" role in the transmission of the disease, as seen in Chang's model.
The model developed by the researcher includes 553,000 different locations that are grouped into 20 categories. All have been visited regularly, according to the model, with residence time and occupation density well represented. The model also "accurately" predicts confirmed daily case counts in ten of the largest metropolitan areas such as Chicago, New York and San Francisco. For example, in the Chicago metropolitan area, the model predicts that 10% of points of interest account for 85% of expected infections.
The model also predicts that lower-income population groups are more likely to be infected because they cannot substantially reduce their mobility, and the places they visit tend to be smaller and more crowded. This would be the case in grocery stores, where up to 59% more people per square meter are detected and customers remain 17% more time, on average.
The model, Chang explains, needs to be complemented by more conventional epidemiological studies both to validate the results and to fill "data gaps", such as the older population or children, who have limited access to mobile phones. Similarly, it allows to determine reopening policies that are more adjusted to the transmission risk. The data is clear on the fact that enclosed, poorly ventilated, high-occupancy places are points of risk. The study does not assess schools or public transport.