After masks and tests, mobile phones to fight the Covid-19 pandemic? On March 26, around 20 researchers from around the world put online a “manifesto” emphasizing the usefulness of telephone data in times of epidemic for ” to alert “, ” Wrestle “, “Control” or “Model”.
Each mobile subscriber, by requesting relay antennas, gives their operator an overview of their trips. The “simple” call listings, after anonymization, can thus make it possible to know how populations move, where there are areas with high density, and therefore at risk, check whether mobility restriction measures are well applied… These techniques have already proven themselves in crisis situations, especially against Ebola. And on April 3, the NGO Flowminder published a preliminary report of mobility analyzes in Ghana, thanks to an agreement with Vodafone, making it possible to estimate the respect of the restrictions imposed in two regions.
Operator data can also improve epidemiological models. These classically consider that the populations are homogeneous, with individuals having the same chances of contaminating each other. The reality is obviously different: there are more contacts at school than in a company, adolescents are more “tactile” … Phones can quantify these interactions in different places, even different age groups. They can also give indications on their evolution between normal and confined period. A subject on which a team from the National Institute of Health and Medical Research (Inserm) is going to work in collaboration with Orange.
But phones can speak more precisely. Researchers and policy makers are seriously considering the use of mobiles to revolutionize the contact tracing, or “contact tracking”. That is to say, tracking, thanks to applications installed on smartphones, sick people and people they are likely to have infected.
Supposed to guarantee the protection of personal data
China, Singapore and South Korea have already taken the plunge. And many other countries are preparing to imitate them, such as the United Kingdom or Germany. In Europe, the device that seems to hold the rope is not exactly the same as in China. Rather than knowing where a patient went, the idea is to identify who the person has been with. And this, without necessarily accessing their movements, but by detecting nearby phones, thanks in particular to Bluetooth wireless technology.
1er April, PEPP-PT, a consortium of European researchers, announced that it is about to launch an IT infrastructure allowing health authorities to build such a patient monitoring application. All the computer code will be open, and the model is supposed to guarantee the protection of personal data. They must allow, they hope, to run applications from different countries together in order to adapt to the movements of populations. The first applications based on this protocol, the last tests of which are in progress, could arrive at the ” half April “. Several governments would follow their work closely.
In the United States, researchers from the Massachusetts Institute of Technology (MIT) are developing a similar application. This would operate in two phases. First, it will be possible for each user to record their movements with GPS and Bluetooth and share them, or not, with a health authority. The latter could, by aggregating the information received, disseminate the risk areas to users. Researchers claim to be working on cryptographic mechanisms making it impossible for the authority to access individual data. In a second step, users could be notified if they have been in close contact with a sick person. This team also prides itself on collaborating with “Many governments around the world” and approaching the World Health Organization (WHO).
“Choice between confinement and telephone contact tracking”
Confidence in this method of contact tracking is based in particular on a study published in Science, March 31, and produced by the University of Oxford. The authors of the report worked on two types of actions supposed to calm the engine of the epidemic (otherwise called reproduction rate, that is to say the number of people that an infected person can contaminate): the effectiveness to isolate the cases and the quarantine of people who have been in contact with a patient.
“The transmission, in the case of the Covid-19, being rapid and intervening before symptoms appear, this implies that the epidemic cannot be contained by the isolation of symptomatic patients alone “, warn the researchers. Hence the idea of also isolating the contacts of an infected person.
This display is old and often used at the beginning of the epidemic to control it and to determine the key parameters of the disease. But the technique has its limits, because it requires filling out questionnaires and field surveys to retrace paths and social interactions.
British Oxford researchers estimate that it should be done with at least 50% efficiency, if not 80%, to rapidly decline the epidemic. This is impossible with the usual contact tracking methods. Only a smartphone application would meet the criteria of quantity and speed. “The choice lies between confinement and this follow-up of contact by telephone”sums up Christophe Fraser, the team leader.
Some researchers also believe that these applications could be useful when deconfining populations to avoid an epidemic outbreak on this occasion. “Rather than quarantining entire populations, we could only do so with those for whom it is necessary. The only way to do this is digitally, “ affirmed, during the presentation of the PEPP-PT project, Marcel Salathé, director of the digital epidemiology department of the Federal Polytechnic School of Lausanne.
While these applications are of great epidemiological interest on paper, no one has ever attempted to develop one for an entire country in just a few days. So far, only localized initiatives, with certainly promising results, have been tried. “My colleagues and I believe that a large-scale electronic contact tracking solution can work if considerable efforts are made to adapt its functioning to existing health processes, and if it is adapted to its users”, says Doctor Lisa O. Danquah of Imperial College London School of Public Health.
There are many limits to this type of application. First, we don't know everything about SARS-CoV-2: for how long is a patient asymptomatic and contagious? On surfaces, from what ” amount “ of virus does the risk of contamination appear? How far and for how long is it considered that there has been risky contact?
“It is not a magic wand”
The number of false alarms and the degree of congestion at the screening locations will depend on the configuration of the system. “These apps are useful, but it's not a magic wand. This can be part of a range of measures. It seems that masks also have an effect, for example, on propagation ”, recalls Alain Barrat, a physicist at the Center for Theoretical Physics in Marseille, who has worked with short-range sensors in schools and hospitals to identify precise interactions.
Nor is it certain that Bluetooth is capable of finely assessing the distance between individuals. The developers of the Singapore application say that for optimal use, the application should be open all the time.
By definition, these applications will only work if they are installed by a significant number of individuals. The corollary, as noted by Michael Parker, professor of bioethics at the University of Oxford and co-author of the article by Science, is that users have confidence in the system.
The concept of anonymous data is misleading
For this, he recommends the transparency of the computer code and its independent evaluation, setting up a supervisory board with citizen participation, sharing knowledge with other countries, etc. “The fact that people are free to choose and not install the application is also a safeguard”, he adds. A survey carried out on March 26 and 27 by his team shows that 80% of French respondents would be ready to install such an application. A survey which has its limits, the respondents having spoken only on the application imagined by the researchers, a priori little greedy in personal data.
This type of monitoring device, on the scale of an entire population, precisely raises the question of personal information and its use by States. Even if the device is not based on geolocation and this data remains on the phone, other information could, in fact, be collected. And the question of the security of the application code – a loophole would allow hackers to steal data – is entirely open.
Whatever the technical solution, these devices will pick up very sensitive data. However, scientists have largely proven that the concept of anonymous data is misleading. Certainly, several experts believe that these applications are not condemned to install mass surveillance. But they still have to be the subject of careful and verified IT development, using proven algorithms. All with the establishment of robust technical and legal safeguards. “It is possible to develop a fully functional application that protects privacy. There is no choice between the contact tracing and privacy. There can be a very good balance between the two, “ says Yves-Alexandre De Montjoye, a recognized expert who heads the Computational Privacy Group at Imperial College London. Provided that we have the means.