Tracking COVID-19 carriers through their smartphones sounds like a smart idea. The hard part is making it work.
Engineers from as far away as Switzerland and Singapore tackled the challenges at a virtual conference on Thursday, sponsored by the Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Laboratory.
They’re trying to use the short-range Bluetooth radios inside the smartphones of COVID-infected people as an easy way to find other people they may have infected. People who’d contracted the virus would agree to run a smartphone app that would send out an anonymous Bluetooth signal to all nearby phones. This signal would indicate to others that an infected person was nearby, and suggest that they get themselves tested for COVID-19.
Dr. KJ Seung, chief of policy for Massachusetts COVID-19 response at medical care nonprofit Partners In Health, said smartphone-based tracking could monitor the majority of people who can afford phones. That way, public health workers could focus on reaching low-income and elderly people who usually must be tracked down in person.
“We want you to deal with the 90 percent of the easy ones," said Seung, "so we can deal with the 10 percent, the tough ones.”
Governor Charlie Baker addressed the conference Thursday, saying in a prerecorded remark that contact tracing efforts are vital "to give people confidence and comfort over time that we can in fact safely go back to work.”
But there are plenty of difficulties in making the technology trustworthy and reliable. Engineers must find ways to make the systems reliably identify infected people and their contacts without also detecting lots of “false positives.” And they’ve got to assuage public fears the tracking apps are not a threat to privacy.
Bluetooth-based tracking systems are under development at a host of academic institutions, including MIT, Boston University, and Carnegie Mellon University. In addition, tech giants Apple and Google have partnered on their own Bluetooth COVID-19 tracker.
Ran Canetti, a Boston University computer science professor, noted that when Google and Apple announced their Bluetooth tracking plan, “immediately we see these headlines about, here is yet another way in which Google and Apple are going to track us all. One needs to be really careful about how to mitigate that, because this is exactly not what we want.”
Canetti said any phone-based tracking system must be based on “open source” software that is available for inspection by independent experts to ensure the absence of bugs or “back doors” that could threaten privacy.
Also, because Bluetooth signals can carry for dozens of feet, a user might be warned of a “nearby” infected person who might actually be 20 or 30 feet away, posing no risk.
But Carnegie Mellon University engineer Swarun Kumar said that an app could use other smartphone sensors to estimate the likely distance to an infected person. Some phones have proximity sensors that know when the device is inside a user’s pocket. When that happens, the user’s clothing will reduce the signal strength of an incoming Bluetooth message. The app can use this information to recalculate the likely distance to the infected person and decide whether he’s close enough to really pose a health risk.
One of the most crucial issues was never addressed: The lack of COVID-19 testing. So far, only about 1 percent of US citizens have been tested, which means that thousands, millions maybe, of people might be unknown carriers of the virus. Until a much larger percentage of the population has undergone testing, contact tracing efforts will be of little benefit, whether they’re conducted by humans or smartphones.