From accessories which you attach to your mattress to smart sensor-studded headphones, there is no shortage of devices out there which promise to monitor (and thereby help improve) your sleep in some way. A new project created by researchers at the Massachusetts Institute of Technology and Massachusetts General Hospital aims to assist the percentage of the population who suffer from some form of sleep disorder. The total number of people affected by sleep disorders in the United States is around 50 million — including those who suffer from diseases such as Alzheimer’s and Parkinson’s which can disrupt our ability to catch some well-earned shut-eye. The difference between this and other approaches? That this project involves no physical contact with users.
Instead, it builds on previous work carried out by MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), showing that a device capable of emitting and receiving lower-power radio frequency (RF) signals can remotely measure a person’s vital signs. This is achieved by analyzing the frequency of these waves, which change slightly as they reflect off the body and can reveal information such as pulse and breathing rate.
The researchers have now built a smart WiFi-like box, which sits in your room and uses these body signature insights to track your sleep, via some deep-learning neural networks.
“This work uses wireless signals and advanced AI algorithms to know when you are dreaming, when your brain is consolidating memory, and more generally, your sleep stages,” Mingmin Zhao, a Ph.D. student who worked on the project, told Digital Trends.
In tests involving 25 healthy volunteers, the researchers found that the technology was 80 percent accurate, a number that is comparable to the accuracy of ratings determined by sleep specialists based on more invasive EEG sensors. Next up, the researchers hope to use the technology to explore how certain neurological diseases affect sleep.
“We are currently working with medical doctors to understand diseases and track response to treatments with this device,” Zhao continued. “There is definitely value in commercialization of the system. We are interested in doing so.”
The work was presented Wednesday at the International Conference on Machine Learning in Sydney, Australia. A paper describing the project is also available online.