Contact tracing has long been used to identify people who have been near someone infected with a transmissible disease. But by the time you’ve been contact-traced, it’s too late for you to do anything to protect yourself. You can’t turn back time and “undo” the interaction you’ve already had with the infected person. You weren’t even told who the infected person was.
Instead, you are asked to comply with precautions – to isolate and inconvenience yourself – to avoid spreading the disease. You do this to protect others, but there’s nothing necessarily in it for you.
So how can we prevent the spread of infections using people’s desire to protect themselves? Po-Shen Loh, a professor of mathematics at Carnegie Mellon University in Pittsburgh, Pa., suggests that you simply turn contact tracing on its head.
Loh specializes in graph theory. In mathematics, a graph is simply a network: a collection of points connected by lines. Human interactions can be represented by a graph. Each person is a point in the graph and two points are connected if the people they represent were in close contact with one another.
Regular contact tracing identifies only an infected person’s immediate contacts. But, Loh asks, why not go beyond that and see just how far away the nearest infected person is? Is it four lines – that is, four contacts, or the contact of a contact of a contact of a contact – away? Twelve lines? Just two? The closer the infection is to you, or the faster it is moving toward you, the greater your incentive will be to protect yourself from the disease, says Loh.
Imagine a disease more lethal than COVID-19. “Would the average person find it useful to get a heads up of: ‘Somebody just got diagnosed with a disease that kills a third of the people who get it. They frequently spend time with someone who frequently spends time with you.’” Loh says that statement alone would make people avoid associating with those they would usually see.
Smartphones have made it possible for Loh to turn this theory into practice with an app he and his team at Carnegie Mellon invented, NOVID. Using Bluetooth, smartphones that have downloaded NOVID can determine if they have been in close contact with each other; while the parameters can be adjusted, NOVID currently defines “close contact” as being in proximity for at least 15 minutes over two days.
People no longer need to go through their memory to list whom they spent time with. A graph encompassing the entire community can be built, with lines between points if and only if the smartphones they represent were in close contact. Some points may have lots of lines emanating from them – these are often hubs of contact, such as servers at restaurants and bars. Other points, such as those representing people sheltering at home, might be connected to no lines at all. All of this can be done without a human contact tracer asking for information.
Manual contact tracing, on the other hand, does not scale well, even to just second-degree contacts. “If each person has 20 contacts and each of these contacts has another 20 contacts, you have to contact 400 individuals, which is doable only if you have … very small epidemics and very good contact-tracing systems,” says Luca Ferretti, a senior researcher at the Big Data Institute at the University of Oxford.
Bar charts in NOVID show how many contacts are infected at each “contact distance” – that is, how many first-, second- and third-degree contacts are infected (and so forth). Those who are infected can self-report, updating the bar charts in real time.
Users might isolate or use personal protective equipment if the disease gets closer to them, says Loh.
Staying safe thus turns into a game of making sure that those bars never reach you – something that Ferretti says he thinks will appeal to users’ curiosity. And if people do get infected, they could get diagnosed and treated early.
Privacy is a major concern with all contact-tracing apps, but NOVID collects no personal information. Users do not need to register their names or phone numbers; their smartphones are represented by an anonymous code. And by using Bluetooth instead of GPS to determine distances, no precise location data is collected.
Loh says that once people learn about NOVID, they would have an incentive to download it, as it would be in their own self-interest to do so. Governments would therefore not need to mandate NOVID, but rather just ensure that their citizens know about it. NOVID “directly help(s) you survive a pandemic.”
How can we prevent the spread of infections using people’s desire to protect themselves?
Ashleigh Tuite, assistant professor of epidemiology at the University of Toronto, agrees in theory but says that because NOVID hasn’t been tested, it isn’t certain that people would download it. And, as with any contact-tracing app, NOVID relies on downloads.
“If there’s not a lot of buy-in and you’re not seeing alerts or information on the app suggesting that people close to you are infected, you may interpret that as being no risk … How do you avoid interpreting the absence of a signal as there being no COVID in your network?”
There’s only one way to find out, says Loh.
“We should test it (NOVID),” he says. “It will fail the first time (perhaps because the user interface is not easy enough to understand or because the parameters about what constitutes a ‘close contact’ are not correctly calibrated), and we should use the failure and learn from it until we make this new concept, that theoretically works, into something that practically works.”
To date, NOVID has been implemented on a handful of university campuses and in one city, Santa Fe, N.M.
“We knew we wouldn’t get 100 per cent adoption just because some people don’t want to put apps on their phones,” says Rich Brown, director of economic development for the City of Santa Fe. “We were an early adopter. We’re at (a vaccination rate of) 78, 80 per cent. NOVID was part of us rising to that number – it was part of the toolkit.
“(But adoption) has plateaued because, I think, people believe that once they’re vaccinated, they’re totally immune. That’s a whole other discussion, but once we reach that high immunity rate, people felt that they didn’t need NOVID to warn them.”
“It’s hard to convince people to use something if they actually see that their chances of dying or being seriously impacted are not high enough. The university students I’m surrounded by – well, after vaccinations, they don’t feel this drive to get onto the system,” says Loh.
Tuite sees applications beyond COVID. “Contact tracing is about trying to understand networks and how people interact with each other and using that information to try to stop disease transmission. There’s nothing special about COVID.”
She adds that researchers have been interested in, for example, using the networks in anonymous dating apps to follow the spread of sexually transmitted diseases.
Privacy concerns notwithstanding, the availability of this detailed information excites Ferretti.
“The data would allow epidemiologists to have first-hand knowledge of how viruses spread,” he says. “What is the typical transmission time of the Delta variant? We barely know that. It looks like we have a lot of data but the good-quality data from which we can learn something – not so much of that.”
NOVID came too late to make a dent in the COVID pandemic. As an app developed independently of both Apple and Google, NOVID could not send out Bluetooth signals if it were only active in the background of a smartphone. Researchers didn’t find a way around this until late in 2020; by that time, the focus of the effort against COVID had shifted to mass vaccinations.
But by the next pandemic, Loh says he hopes that everyone will know about the concept behind NOVID – and that it too will go viral.
The cover image is courtesy of the developers of the NOVID app.