On the trail of the virus: How are Corona apps used throughout the world, what technical difficulties are encountered during their implementation and in which ways are they currently being refined. This is a comparison of Corona apps from different countries.
With the Corona pandemic’s outbreak, a call has quickly gone out around the world for technological tools to contain the virus’ spread. Ahead of one global solution, many countries around the world are using their own technology-based variations, so that within just a matter of months a heterogeneous landscape of smartphone apps has emerged.
In broad terms, three groups of Corona apps can be identified
These allow people to have their symptoms assessed, before receiving guidelines for follow-up action. This advice might be to do with virus testing or even going into quarantine. Such apps are being used, for example, in Spain.
Apps for monitoring quarantine restrictions
A great many people are carrying around small computers in the shape of their smartphones virtually all the time and these can also be used for surveillance. Location data tells you whether or not people are complying with quarantine restrictions.
In several countries – Poland for example – there is mandatory use of such an app, monitoring locations via GPS data. Such checks take place, for example, either by asking users to send selfies with locations activated or by making unannounced video calls. In Moscow, public surveillance cameras in connection with facial recognition software are also used to punish those violating restrictions.
In Turkey, all over 64-year-olds must use a particular app that constantly logs their whereabouts and asks them to go back home if they stray too far. People entering Hong Kong must wear wristbands during quarantine to track any change in their location.
Contact tracing apps
This third group of apps help to trace infection chains. The term contact tracing, which is commonly used for these applications, implies locating infections so as to prevent the virus from spreading further. In the past, public health offices have pursued this task only analogously. However, in the case of the Corona pandemic employees are now given digital helpers with these apps.
The functioning of contact tracing apps
Contact tracing apps can be based on two technical approaches: the location determination of an individual mobile phone using GPS; or a measurement of the distance to other devices using Bluetooth Low Energy (BLE).
Both variants have their downsides. Measuring GPS signals from buildings is difficult so that the app fails in exactly those situations where transmission of the virus is more likely. The use of Bluetooth Low Energy also has technical limitations. For example, researchers have found limited efficiency in public transport (trams or buses). Furthermore, Bluetooth Low Energy does not work smoothly if the app using it is not active in the foreground. This was made possible only in mid-May as a result of changes made by Apple and Google at the level of the iOS and Android operating system by providing an interface for contact capturing using Bluetooth Low Energy.
A further distinction is made by the type of data storage, in that it can be centralised or decentralised. The former’s advantage is its continuing availability for use in further epidemiological analysis (as in France and Norway; see country descriptions below). The latter, where information is stored only on individual smartphones, meets strict data protection requirements and is thus intended to promote trust in voluntary use. This is required by Apple and Google for the use of their interface.
Germany uses such a contact tracing app for back-tracking in its ‘Corona-Warn-App’. What follows, therefore, focuses on such applications.
The efficacy of contact tracing apps
Contact tracing apps are subject to the network effect, meaning that the more people use them, the greater will be the benefit for all. Hence, the decisive factor rests not only with how many people constantly carry devices, but also all devices’ ability to record encounters with each other. In the event of a device user’s reported infection, this information is passed via the app in retrospect. Besides the purely technical functioning of software, this also requires a technical connection to laboratory assistants who in turn transmit detected infections to those affected as quickly as possible.
In most countries, the use of such an app is voluntary (exceptions include, for example, Singapore and India, see below).
According to a study by Oxford University, the theoretical limit for such apps to be the only measure that can stop the pandemic is a distribution to 60% of the population. However, the study’s authors also stress that even much smaller levels of distribution can positively affect the number of infections. To put this into perspective, around 80% of Germans aged 14 years and older own a smartphone. Hence, the necessary distribution threshold would theoretically be achievable, albeit up-to-date operating systems would also need to be installed in order for the app to be totally effective.
The upper limit of an app’s range can be estimated by the number of downloads, although it should be noted that this number does not necessarily equate to the number of active users. For example, it could be that the app has been downloaded to a mobile phone, but the Bluetooth function is not activated, or that users delete the app and then download it again. Accordingly, the actual spread will always be below the download figures.
The true value of contact tracing apps on the level of infections has yet to be proved. For instance, a review study published in The Lancet Digital Health for August 2020, reported that “no empirical evidence of the effectiveness of automated contact tracing (regarding contacts identified or transmission reduction) was identified.”
How contact tracing apps have spread internationally
There now follows a description of contact tracing apps which have been introduced in Singapore, Austria, India, Norway, France, Germany and Ireland with a comparison of the different apps being used. From a time perspective, three phases of introduction can generally be identified:
Phase 1: February to April
For decades, the epidemiological method of contact tracing has been established to track and interrupt infection chains. Digital technologies are now available which can complement this traditional arrangement: With apps on smartphones, contact tracing can be automated and integrated into existing healthcare system processes.
In Europe, international research groups were set up in the Spring of 2020 to develop appropriate concepts. Debates about the concrete design have been focused mainly on data storage: should it be centralised or decentralised? Should this app work on the basis of location data or should it analyse the encounter of people or their mobile phones via Bluetooth Low Energy radio technology? The crucial question behind this is: what is the relationship between public health protection and the right to privacy?
In countries such as Singapore or Austria that went live early, apps are being released that struggle with the technical limitations of Apple and Google operating systems (see below).
Phase 2: April and May
In many countries, debates over the technical design ended on 10 April, 2020. On this day, Apple and Google announced their collaboration to fight the pandemic. The vast majority of all smartphones run on one of these firms’ operating systems. The so-called exposure notification interface overcomes previously existing technical limitations in the use of Bluetooth Low Energy. Apps can now also register contacts when they are running in the background, rather than when they are active just in the foreground. This interface allows only decentralised storage. Since 20 May 20, it has been available to a public institution in each country for use with an official app.
Phase 3: June to September
Especially in Europe, more and more governments have been making their contact tracing apps available for download. According to a progress report by the EU Commission in June, most are based on the interface between Apple and Google. Only France (see below), Hungary and Slovakia rely on centralised data storage. In this case the interface with Apple and Google cannot be used.
The parallel development of apps in different countries is leading to difficulties. National apps, whether they use centralised or decentralised data storage, are not compatible with each other. This means, for example, that a person’s mobile phone with an app from a German app store cannot record contacts to a phone running an application from an Italian app store while on holiday in Italy. This linking function, though, is to be implemented during the Autumn (see below).
As early as March this year, Singapore was one of the first countries to introduce a tracing app, but problems quickly became apparent. The application in Singapore is based on Bluetooth technology and not able to rely on the interface introduced later by Google and Apple. This had several technical disadvantages: the battery power is quickly used up; the app has to run in the foreground; and it is not possible to use other Bluetooth devices such as headphones at the same time. Hence, the download rate was relatively low, especially in the beginning, but it is now more than 35% of the population.
This is also due to the fact that use is compulsory for certain demographic groups, such as foreign workers living in collective accommodation.
The country has also developed a so-called Bluetooth token. The device can take over the same task as the app, but does not depend on a smartphone. This is intended to benefit in particular those people who do not use a smartphone.
In addition to the tracing app, the government uses another app to monitor mandatory periods of quarantine with selfies or unannounced video calls.
By the end of March, and thus only a little later than Singapore, the Austrian Red Cross released its ‘Stop Corona App’. This was developed by the IT service organisation Accenture and is based on Bluetooth distance measurement. The costs of initially around two million Euros were borne by a private trust fund. The launch was accompanied by a public debate on data protection and voluntary use.
Similar to the app in Singapore, the Austrian application had technical difficulties. These resulted, for example, due to unreliable recording of contacts because of technical restrictions, there being no Apple/Google solution existing at that time. A major shortcoming turned out to be the ‘manual handshake’ on Apple devices. The storage of contacts did not take place automatically, but each encounter had to be confirmed by users in the app. This conflict situation of unfinished political debates and technical issues naturally led to the app being given a poor image.
Once a switch to the interface developed by Apple and Google had been achieved, the technical deficits were largely eliminated by June. The download numbers have nevertheless remained relatively low, so that nationwide by the end of August the app had been downloaded only around 900,000 times.
The Indian Federal Government’s contact tracing app ‘Aarogya Setu’ is available in eleven languages and based on a previous app that was released in April. In addition to the national app, there are several other official apps available from individual federal states. The national app combines Bluetooth technology for contact tracing with GPS location tracking. This enables warnings to be sent to users, for instance when they are in risk areas. Centralised data storage applies.
The use of the app is obligatory for large parts of the population in India, the only democratic state so far to have taken this extreme step. Both government employees and those working in private companies must all install and use the app. Additionally, use of the app is to be made mandatory in particularly hard-hit areas. However, only about a third of the Indian people own smartphones. The obligation to install this app is controversial, mainly due to its lack of a legal basis.
India’s large population has given the tracing app a ‘superlative’ on absolute figures: no other smartphone app in the world has ever reached 50 million downloads in less time. Hitherto the record was held by the game Pokémon Go. Furthermore, by September downloads had risen to more than 150 million.
The Norwegian contact tracing app ‘Smittestopp’ was launched comparatively early in mid-April, but has since been discontinued after major data protection concerns. In contrast to Germany’s ‘Corona-Warn-App’, the Norwegian app used GPS as well as Bluetooth to track: where the devices are; and which other mobile phones are in their vicinity. The analysis of the movement profiles was intended to help trace the virus’ spread and government efficiency levels.
This app was developed by the Simula Research Laboratory on behalf of the Norwegian Public Health Office, FHI, within five weeks. The Norwegian developers decided not to publish the source code, because they were afraid that due to possible weaknesses the protection of personal data could no longer be guaranteed. In addition, they assumed that the advantages of Open Source through collaboration would not work with the app’s already limited runtime, namely only during the pandemic.
The lifetime of the app was then even shorter than expected. In June, after only two months of live activity, it was discontinued following a warning from the Norwegian Data Protection Authority. Thereafter, all data previously collected and centrally stored was deleted. Even before the app was discontinued, though, it had already lost a huge number of users. At the beginning of June, their number was only just over half a million, with more than one and a half million downloads.
Amnesty International also strongly criticised the app and, together with applications from Kuwait and Bahrain, classified it as extremely worrying because users’ whereabouts could be tracked.
In early April, the French government announced its development of a contact tracing app ‘StopCovid’. The app was developed free of charge by a team from public and private organisations led by the National Research Institute for Digital Science and Technology (Inria). The code, released partly as Open Source, is based on distance measurement via Bluetooth and centralised anonymised data storage combined with high security and data protection aspects (the so-called ROBERT scheme; ROBust and privacy-presERving proximity Tracing).
The French government thus deliberately decided against using the Apple/Google interface. This was for two reasons: Firstly, France wants to remain independent of the big US tech firms; and secondly, the country is planning on using the anonymised data for further analysis. This decision had its price. Especially on Apple devices, which account for about 20% of smartphones in France, the app works only to a limited extent. This is largely because the app when used on iPhones should always be active in the foreground. Despite a request from the government to change this, Apple was not willing to do so. The French developers did manage to find a technical diversion where devices with Android operating systems could ‘wake up’ Apple devices and thus at least reduce the problem.
One consequence of the French special path is that a planned EU-wide data exchange is unlikely to work since France’s centralised approach is not compatible with other countries’ decentralised apps.
The download figures for the app, released at the beginning of June, are rather low: by mid-August, it had been downloaded only 2.3 million times. Since then, infections have been reported via 1,169 QR codes and 72 warnings have been issued.
The federal government released the ‘Corona-Warn-App’ during June 2020 in the app-stores of Google and Apple. This was preceded by a debate on the technical design. It essentially revolved around whether the data should be subject to centralised storage on a server or local storage on users’ devices. The original plan of Health Minister Jens Spahn was to store the data centrally. After a public debate and the decision by Google and Apple to provide an interface (API) that allows only decentralised storage, the government switched accordingly.
The Robert Koch Institute regularly releases key figures for the app, reporting that by the beginning of September, it had been downloaded 18 million times. Furthermore, just over 4,300 people received a so-called teleTAN after a positive Corona test, which they could use to warn their contacts via the app. The technical connection of laboratories to the ‘Corona-Warn-App’ remains problematic, with only two thirds of all laboratories being directly connected to the digital logging system. Infected people who have been tested by a laboratory without a direct connection must first obtain the TAN by calling a hotline.
With support from a large-scale federal government campaign, by August the ‘Corona-Warn-App’ was known to 85% of those included in a survey conducted by Initiative d21 and the Technical University of Munich. However, 51% of all respondents believe that the warning app does not bring any benefit at all.
Even though the German Corona tracing app has been active since June, the discussion about possible improvements continues. There is, for example, the proposal to develop the app further in such a way that an accumulation of infections (clusters) can still be detected while maintaining the data-saving approach.
The Corona tracing app in Ireland is also based on the Google and Apple interface. As with the German version, it is considered to be success, as measured by the number of downloads. Within the first 24 hours of its release, the app was installed more than one million times on smartphones, in other words by 20% of the population.
Contrary to the German ‘Corona-Warn-App’, this app was not developed by big firms, but programmed by the software company Nearform which has only 150 employees. It cost the Irish government a mere 850,000 euros.
In the meantime, Northern Ireland and Gibraltar have adopted this solution and Scotland has also engaged the company for early provision of its own tracing app. It is likely that more countries or regions will draw on this technical basis. The Irish Health Service and Nearform released the source code at the end of July as Open Source under the name Covid Green and handed it over to the newly founded Linux Foundation Public Health.
Outlook: What’s next?
It is imperative to note that technology alone will not end the Corona pandemic. Only retrospective analysis will show whether contact tracing apps are a useful additional measure in its containment. The rapid technical implementation is currently a giant field experiment that cannot take place without trial and error, mistakes, setbacks and accompanying socio-political debates.
Corona tracing apps are now approaching an emerging fourth phase, with international systems being developed.
The countries of the African Union are starting an ambitious project with PanaBIOS with the aim of reopening closed borders. Travellers can also use their test results in another country via an app, because results are subject to centralised data storage.
With the EU Federation Gateway Server, the EU wants to link national apps with each other. Member states agreed on the technical specifications for this in June. Following an EU Commission initiative, the European decentralised apps will be made compatible with one another – implemented by Telekom and SAP, who have developed the German tracing app. Apps from France, Hungary and Slovakia, which are based on a fundamentally different architecture (centralised data storage, without Google-Apple-API), will be excluded. In addition, Apple and Google have announced their intention to make the functionality of their interface available in the future also at operating system level. This means that an additional app would then no longer be necessary for contact tracing.
MIT researcher Katharin Tai describes the use of apps in Asian countries and their integration into the general containment strategy.
Fabio Chiusi discusses the European variants in the use of technology. This analysis is complemented by 16 examples of cases from European countries.
Essays and analyses with assessments from over 30 countries on how surveillance technologies have been introduced in response to the Corona pandemic.
Projects that try to follow international technological developments continually:
The blogs published by the bidt represent the views of the authors; they do not reflect the position of the Institute as a whole.