Image: An AI algorithm monitors sleep stages without sensors attached to the body (Photo courtesy of Christine Daniloff / MIT).
Low-power radio waves that detect small changes in body movement caused by the patient's breathing and pulse can nonintrusively diagnose and study sleep problems.
Developed by researchers at the Massachusetts Institute of Technology (MIT, Cambridge, MA, USA) and Massachusetts General Hospital (MGH; Boston, USA), the device uses an advanced artificial intelligence (AI) algorithm to analyze the radio signals surrounding the person. As the radio waves reflect off the body, any slight movement of the body alters their frequency. The AI algorithm analyzes those waves and translates the results into sleep stages: light, deep, or rapid eye movement (REM).
The AI algorithm is based on deep neural networks, which extract and analyze complex datasets in order to isolate frequency measurements and identify them as pulse, breathing rate, and movement, while eliminating irrelevant information. In a study of 25 healthy volunteers, the new technique was found to be about 80% accurate, comparable to accuracy ratings that were based on electroencephalogram (EEG) measurements. The study was presented at the 2017 International Conference on Machine Learning, held during August 2017 in Sydney (Australia).
“The opportunity is very big because we don't understand sleep well, and a high fraction of the population has sleep problems,” said lead author MIT graduate student Mingmin Zhao, MSc. “We have this technology that, if we can make it work, can move us from a world where we do sleep studies once every few months in the sleep lab to continuous sleep studies in the home.”
“Imagine if your Wi-Fi router knows when you are dreaming and can monitor whether you are having enough deep sleep, which is necessary for memory consolidation,” said senior author professor of electrical engineering and computer science Dina Katabi, PhD. “Our vision is developing health sensors that will disappear into the background and capture physiological signals and important health metrics, without asking the user to change her behavior in any way.”
REM and non-REM sleep alternate within one sleep cycle, which lasts about 90 minutes. REM sleep is characterized by rapid random movement of the eyes, dystonia, and vivid dreaming. It is also known as paradoxical sleep (PS) because of physiological similarities to waking states, including rapid, low-voltage desynchronized brain waves. REM sleep causes marked physical changes, including suspended central homeostasis, which allows large fluctuations in respiration, thermoregulation, and circulation, which do not occur in any other modes of sleeping or waking.
Massachusetts Institute of Technology
Massachusetts General Hospital