MIT researchers have been building out the capabilities of a system they created that can monitor vital signs without requiring any direct contact sensors or wearables -- using wireless signals already present in the environment. Now the team, which is working out of MIT's Computer Science and Artificial Intelligence Lab (CSAIL), is taking their system even further, with the ability to identify a specific individual as one they've monitored previously, keeping their records tied to them over time, all while preserving privacy by not connecting this to any personal information about the subject.
The new tech, which is called "RF-ReID" is useful because it could allow for monitoring individuals cohabiting in a group over time, like seniors in a retirement or long-term care facility, for instance. This development is particularly important because the ability to monitor an individual over time is crucial for actually being able to observe and detect any deviation from a healthy baseline.
While healthcare facilities obviously employ a number of different measures to monitor resident and patient vital signs over time, these can run into potentially serious limitations. Cameras aren't very privacy-respecting, and they also have limitations in terms of identifying individuals consistently over time, depending on even superficial changes to their appearance, like wearing different clothes. Remote monitoring devices are only as good as the memories and consistency of the individuals employing them, whereas the MIT CSAIL system works independent of any on-person device.
The team that developed RF-ReID say that It can re-identify a new individual introduced to the system after less than 10 seconds of physical activity, using signals including body size, walking speed and gait style, all of which it infers from data collected by wireless radio signals present in the environment.
Because it doesn't need any existing biographical or personally identifiable private information to work, this could prove the basis of privacy-preserving monitoring systems that could be used even in the identification of potential COVID-19 cases in care homes. The researchers suggest that the system could flag any potential onset of symptoms, prompting care workers to do an in-person round of screening and/or testing.