International use of unique mobile data from DTU

When smartphones began to become everyone's gadget ten years ago, Professor Sune Lehmann got an idea: He wanted to use mobile data in a social networking experiment where people's movements and networks were mapped. Together with his group at DTU and a number of partners at the University of Copenhagen, he designed the project ‘SensibleDTU’.

So when the academic year 2013-14 began, 1000 first-year students were equipped with a free smartphone in exchange for DTU having to use the mobile data in an anonymised way in accordance with all specially regulated regulations, permits and privacy using a specially developed software.

For three years, the researchers collected a huge data set on mobility, which was increased daily by 50-100 GB. At the time, the idea of a pandemic was of more academic than practical interest. But when the world suddenly lacked knowledge for infection detection, infection-stopping apps based on Bluetooth technology, and shutdowns and reopens, the dataset became interesting because no one had done anything similar. In recent weeks, several pandemic articles have been published with the DTU dataset, including in two Nature journals, with Sune Lehmann as co-author.

“I am really proud that a DTU experiment designed in 2012 still proves to play a huge role in science with now two nice publications in which DTU participates. Without having planned it, we suddenly have a data set of measurements that are, first of all, groundbreaking worldwide in relation to the size of the population, but which also turn out to be incredibly useful, because we estimated from Bluetooth how close people were to each other, and where they met in the real world,” says Professor Sune Lehmann.

The professor at DTU Compute is often contacted by foreign researchers and authorities who want to use the data set. He himself participated in the advisory board when Denmark had to develop its own SmitteStop app.

The importance of social networks for Covid-19

The newly published articles are in the magazines Nature Physics and Nature Communications.

The first study led by researchers at Indiana University in the United States shows how to use so-called backward tracing to effectively detect and prevent infection. The DTU data show that some people have a lot of friends (works as a kind of friendship hub), while others do not have as many. Here, the researchers use the data to show that it is also important to find Covid-19 sources of infection with a lot of friends quickly, while they are still contagious in order to stop the infection that way, just like with the newly infected network. In Denmark, both methods are used - backward tracing and forward tracking - to stop the infection, but the article emphasizes the effectiveness of backward tracking.

The second major study, led by researchers from the Italian University of Trento, is about digital contact tracing. The original analysis of digital infection detection is based on a simple model with many assumptions. Here, on the other hand, the researchers look at how to put the most realistic parameters into, for example, infection stop apps, so that infection detection becomes as effective as possible. Here, the unique DTU data set contributes to real data and thus realism.

This week, a not yet peer review-approved article has also been published with knowledge from the data set. In this article, a group of researchers at Harvard University in the United States investigate how to safely reopen a campus after the Covid-19 shutdown. Here, the researchers have used the Bluetooth-based data set from DTU instead of simulated data to investigate and assess different reopening strategies in relation to the spread of the virus. They look at what testing, isolation, facemask and social distance mean.

Timing and unique research

When the data set is being used so much, the timing has a big impact, says Sune Lehmann. In fact, DTU republished the data set from their mobile experiment in an open version in late 2019, shortly before the pandemic.

That the SensibleDTU study with mobile data was unique is also emphasized by the fact that one of Sune Lehmann's old PhD students from the project today works at Waymo, which is part of Google's parent company.

"In connection with the development of Google and Apple's infection detection API, he was involved in the work of developing the algorithms used by the Bluetooth-based contact detection technology in infection control apps on Android phones, where users are still guaranteed a high degree of privacy, ” Sune Lehmann says.