Why there is an app for everything except Covid-19

0
2

                                        I am a cool lover of digital technology. However, I notice that techies and influencers accuse the Belgian government of wasting money because it does not opt ​​for innovative technology to trace corona contacts. I would rather not talk about wasting money, but the idea that technology, and specifically a contact tracing app, is an adequate solution to replace contact tracing call centers is simply wrong. There is no workable contact tracking app because people are more sophisticated than the technology available.                                        </strong>
                                </p><div>
                                                                        <a name="6934640"/>

                                                                                      <p>Let's take the example of TraceTogether, the most talked about Singaporean app, which is also the basis for the Australian Covidsafe app. We assume that many people download and use this app and have their smartphone with them as always. Let's assume iOS and Android are interoperable (the exposure notification api has only recently been released by Apple and Google). Finally, imagine that most people with a positive Covid-19 test notify the authorities through the app. The app registers anonymous connections between two smartphones, the device model, the time and the strength of the signal. Depending on the device, the version, the technology stack and the physical environment (e.g. the building density), Bluetooth can handle a distance between ten and one hundred meters. An algorithm calculates the distance based on the smartphone model and the signal strength. Well, this is not a good tracing solution because of the variable signal strength and the various environmental factors

The problem lies in under- and over-reporting. In the first place, Metcalfe’s law is at play. Suppose half of the population uses the app, a quarter of the contacts are reported. Suppose that eighty percent of the population has installed the app (that’s just about every American with a smartphone, an obligation), the number of reported contacts rises to just 64 percent. With ten percent adoption, you end up with negligible one percent reporting. In Singapore, barely a quarter of the population had downloaded the app in mid-May, despite the political culture and incentives that also included shopping discounts. That means six percent (!) Reported contacts. So it’s all about the reported contacts versus the real encounters. In more or less closed settings with many interactions (such as a busy bar), there are many contacts, but the percentage of reported contacts remains the same. In such environments or in a fairly closed ecosystem, the higher number of contacts (in a bar with a hundred people, for example, the number of unique contacts can be up to 4,950), which may lead to a larger application, but it always remains the same as the adoption percentage. And then we assume that the phone always goes with the owner. That may not be the case on a beach. Under-reporting also stems from the app that tracks the phone and not the virus. A person’s interactions with the physical world, through door handles, on toilets, and on trains, do not create reportable contacts.

                                                                                                                                                                                                    <h3>Over-reporting</h3>
                                                                                                                                 <p>Over-reporting is another problem. For example, last weekend, two days of shopping, walking, driving and cycling, means that I was in close contact with a thousand people. Okay, that number can vary from person to person, but unless you're a couch potato that has stored canned food and water in your nuclear bunker for seven years, the number will be high. Bluetooth is about walls, glass, store shelves and safety screens. It does not understand the safety signs in streets and shops. It does not know whether the user of the app is in the car or wearing a mask. It does not know who cuddles, sneezes or coughs. There is no context. As a contact tracer calling me and asking questions about which close contacts I had over the past three weeks, it is easy for me, as an introvert. These are at most a handful of meetings that I will undoubtedly remember. For the tracers, there is so much data noise in the app that it's probably better not to use it at all.

Now, let’s say the tracers would use the bluetooth data and contact everyone on the reported list. Can you imagine how much fear would arise? I should know because in March I received such a call from a travel agency that someone stayed at my hotel who tested positive four days later. And do you put yourself in quarantine? Are you putting a group or region in lockdown with potentially incorrect data? At best, the data is a tool, but probably a distraction with terrible side effects.

                                                                                                                                                                                                    <h3>Freewheeling</h3>
                                                                                                                                 <p>For the good of the digital fans, let's go a little further and freewheel a bit. Let's add location data to accurately calculate the distance even though GPS is struggling with walls including the levels of buildings (ever been to Singapore?). Yes oops, there goes your privacy but maybe you can live with that because the data is only in your smartphone, except for the confirmed cases. Here we will say the higher goal. Let's assume that all our devices are from after 2018, have the correct GPS receivers and use the L5 band. This is necessary to register a location accurately enough (in open spaces). To avoid over-reporting, we need more contextual data and artificial intelligence (ai). Location alone is insufficient. You have to map movement and speed, if only to distinguish between walking, cycling or driving. You can record this on your smartphone as Garmin or Strava does. However, you cannot do that for your contacts while it is necessary to indicate the relevant meetings. So let's get away from the idea of ​​only storing data on the smartphone as long as no positive test is indicated. Then let's put all the data of all people and all movements in the cloud. We can let go of that. We can look at other data sources to make AI smarter, for example to really profile people. Then we need to address the important underreporting. The mandatory use of the app is really a sine qua non. Everything could be better: the smartphone may not be the best instrument.

Why not a device (hello, chip) that we wear permanently? Telco’s are no doubt drooling about the possibilities of iot and 5G. The sky is the limit. Why use these applications only for Covid-19 and not for other possible solutions? Those who want to put us on that path would best quarantine for a few weeks, look in the mirror and think deeply. I, and the vast majority with me, will never accept a society under such supervision. The goal should not be an excuse for dangerous substances.

There is good reason to choose contact tracing centers. People in call centers will, for a long time, be better at asking the right questions than any app.

Look, the best approach is still discipline, that everyone adheres to the rules and adheres to the guidelines. This offers good quality guarantees. Let us also hope that research institutions and other organizations, in order to combat the pandemic, can provide sufficiently accurate tests, vaccines and medicines.

Herman De Prins is CIO of UCB, ‘Cio of the year 2016’ and ‘European Digital Innovator of the year 2017’

            </div>

LEAVE A REPLY

Please enter your comment!
Please enter your name here