MIT has an ambitious plan to detect if you’ve been exposed to the coronavirus

There are three basic steps to stamping out the spread of a highly contagious disease such as COVID-19. First, you detect who has the disease. Then, you trace who they may have passed it to. And finally, you isolate everyone who is at risk.

Step two—also known as contact tracing—can be a time-consuming, manual process where public health officials ask those who’ve fallen ill to create lists of people they’ve been in contact with so those people can be told to self-quarantine. In South Korea, the government has turned to technology to scale up this process, using cellphone location data, CCTV cameras, credit card purchase data, and other kinds of surveillance to tell people if they’ve come into contact with individuals who have tested positive for the virus.

“That contributed to an extremely successful comprehensive approach in controlling the outbreak in South Korea,” says Louise Ivers, an infectious disease expert, associate professor at Harvard Medical School, and executive director of the Massachusetts General Hospital Center for Global Health.

Effective though it may be at helping control COVID-19’s spread, giving the government full access to people’s locations violates privacy and undermines civil liberties. That’s why Ivers has joined up with experts at MIT to advise on the creation of an automated contact tracing system that can help public health officials and individuals more effectively combat COVID-19—without putting people’s private information at risk.

Instead of relying on GPS-based location data extracted from millions of people’s cellphones, the MIT researchers have built a system that uses random identifiers transmitted via Bluetooth—the same technology that you use to connect your smartphone to your headphones. Bluetooth signals also have another benefit: They’re just more accurate than GPS, which often doesn’t function inside buildings.

“Instead of an eye in the sky that watches everybody, we want to have the phones that people are carrying around tell how close they’ve been to other people,” says Ron Rivest, an Institute Professor at MIT Institute and the project’s principal investigator.

Rivest’s team has designed a system, called Private Automated Contact Tracing (PACT), where your phone would constantly emit a random string of numbers, like an anonymous ID, via Bluetooth signal. Your phone would also keep a running list of any Bluetooth signals and their associated numbers that it detects within a certain set of parameters—such as within 6 or 7 feet, for a duration of 10 minutes or more.

When people test positive for the virus, a public health official would give them a QR code to scan in the app, which would trigger the upload of their entire Bluetooth ID log to the cloud. Then, anyone who has the app on their phone would receive a notification if they happened to be in close contact with the infected person—which could encourage them to enter into self-quarantine.

“We’re fundamentally keeping track of . . . relative proximity, versus everyone’s location,” says Daniel Weitzner, a principal research scientist in the MIT Computer Science and Artificial Intelligence Laboratory and co-principal investigator of the project. “We can hone in on the contact events that are medically relevant without having to track everyone, where they are, all the time.”

So far, members of MIT’s Lincoln Lab have been testing out a prototype of PACT on about 100 devices. That’s helped the researchers work out kinks—such as making sure that the system works on both iPhones and Android phones. There are a lot of challenges to a system like this: For instance, the orientation and angle of a phone, along with its physical location (like if it’s located in a purse, backpack, or pocket), can have a significant impact on the strength of the signal. The researchers also haven’t been able to verify 100% accuracy by testing how the signals transmit in different environments, like a forest or a supermarket. Nor have they been able to directly measure exactly how far away people are from each other, given that everyone who is part of the project is isolated at home with their families.

“We understand there’s going to be imperfections in this approach, but we’re still optimistic,” says Mark Zissman, the associate head of the MIT Lincoln Lab’s cyber security and information science division. “We’re doing the modeling right now to show people that even with imperfections this will still bring some value to the community.”

In addition, some people don’t have phones, which is why the researchers believe that PACT should function as a supplement to a bigger, manual contact tracing program—like the one that Massachusetts governor Charlie Baker announced last week. Integrating the automated Bluetooth technology into a public program like this will require the help and expertise of public health officials. Ultimately, that may manifest as a dashboard that links both manual and automated contact tracing efforts. “What that dashboard looks like depends on public health authorities that need to lead the entire endeavor,” Ivers says.

While Lincoln Lab has built a working prototype, Zissman believes that the MIT team shouldn’t be the one to code the final app. “It needs to be done in my mind by a very experienced professional app developer team,” Zissman says. “For iPhones, what would be better than getting Apple to do that, and for Android, what could be better than Google?”

The team says it’s already in talks with high-level officials in the Massachusetts state government, and they’ve also begun discussing the idea with Apple and Google. But even if both tech giants come on board to build the PACT app, it will also require mass public adoption. To be effective, an estimated 60% of the U.S. population needs to opt in. The team admits that figuring out how to convince Americans to download and use the app is an open question, though it plans to work with public officials to do so. Apple in particular could also play a role by including the app in a software update and pushing it out to millions of phones.

All these challenges are a daunting proposition for a research project. And without a clear sense of when the PACT app might get built or how it will be integrated into the larger public health effort, it might be too late to make a substantial impact on the coronavirus’s spread—though it could help mitigate a future spike in COVID-19 cases as well as other pandemics.

But Ivers believes that despite these obstacles, the visionary scope of the project matches the scale of the challenge at hand.

“We are all staying at home. Our economy is going down and down. People are losing jobs. We have to have an ambitious plan to get out of this,” she says. “If we don’t think big, we’re going to have a lot of people get sick and a lot of people die.”