MIT’s Safe Paths Aims to Inform Individuals About COVID-19 Where They Live

Although the COVID-19 pandemic is by definition a worldwide crisis, it is still distributed in a very skewed manner–and it would help us to know better where it is. A team run out of MIT is working on this at a very detailed, grass roots level though an interesting system called Safe Paths. Although public health authorities and clinicians want information that will help them apportion resources and recommend actions, Safe Paths is directed toward individuals. It lets you ask: Have I recently crossed paths with an infected individual? Was somebody in the grocery store or post office around the same time I was?

Naturally, the best way to save lives is to minimize all physical contact. Those trips to public places should be done only for critical needs. But most of us have been lax until recently, and many (although no one reading this article, I trust) are still being lax. The Safe Paths developers wants to help us look back in time and assess just how much danger we have been in. Their rationale and method are discussed in a white paper.

The process begins with a person who is diagnosed with the disease or learns that she should self-isolate because she was in contact with a carrier. At that point, she should download an app called Private Kit from MIT, available on IoS and Google Pixel. The app creates a history of the individual’s movements, based on the times and locations of the phone over a period of days or weeks.

Next, the individual shares her history with a central server, with no identifying information.

Finally, other individuals download information from the central server and find out any overlap between their locations and those of the potential carriers.

The process is technically simple and preserves the anonymity of all parties. The Safe Path developers are exploring the possibility of peer-to-peer data exchange, but that’s a speculative project for the future.

The quality of the information we get depends on how many infected people supply their information. If the system is judged useful, it would be great to ask health officials to recommend the app as soon as they deliver the bad news to individuals who have been exposed to the virus. We still have to wonder how many exposed people actually are carriers, particularly in the many countries–including the U.S.–that are slow to test people. But perhaps we’re at an early enough stage in contagion that we can learn a lot simply by tracking everyone who knew they were exposed.

There are many other programs for tracking the spread of COVID-19, of course. Several countries collect location data on their residents without asking permission and without concern for privacy. This has been legitimately criticized as a step toward general, authoritarian government surveillance.

Kinsa Health, which successfully improved predictions about influenza outbreaks last year, is applying its same data from Internet-connected thermometers to highlight regions where there’s likely to be COVID-19 infections. Unfortunately, COVID-19 is critically different from influenze, in that carriers are most infectious before they show symptoms. The Kinsa Health data will probably still be useful for COVID-19, but will be less tightly linked to actual outbreaks than it is for the flu. And the data is aggregated over geographic areas, whereas the Safe Paths data is extremely local.

Another effort to gather location data comes from Sentinel Healthcare, which requires a specialized device.

One of the benefits of programs like these will be to reduce panic, according to Kaushal Jain, a member of the Safe Paths team I talked to. If people discover that they had few or no contacts with carriers (that is, among the people who chose to share their information), they can feel better. People who were exposed have an incentive to use the app in order to protect friends and family, which may help spur adoption.

About the author

Andy Oram

Andy Oram

Andy Oram writes and edits documents about many aspects of computing, ranging in size from blog postings to full-length books. Topics cover a wide range of computer technologies: data science and machine learning, programming languages, Web performance, Internet of Things, databases, free and open source software, and more. My editorial output at O'Reilly Media included the first books ever published commercially in the United States on Linux, the 2001 title Peer-to-Peer (frequently cited in connection with those technologies), and the 2007 title Beautiful Code. He is a regular correspondent on health IT and health policy for He also contributes to other publications about policy issues related to the Internet and about trends affecting technical innovation and its effects on society. Print publications where his work has appeared include The Economist, Communications of the ACM, Copyright World, the Journal of Information Technology & Politics, Vanguardia Dossier, and Internet Law and Business.

1 Comment

  • This is pretty genius. Google already knows everywhere I’ve been and should be able to tell me if my path has crossed with someone with a known infection. I hope this scales up.

Click here to post a comment