Catching a breath -- wirelessly
University of Utah engineers who built wireless networks that see through walls now are aiming the technology at a new goal: noninvasively measuring the breathing of surgery patients, adults with sleep apnea and babies at risk of sudden infant death syndrome (SIDS).
Study authors estimates it will be five years until such a product is on the market. They say a network of wireless transceivers around a bed can measure breathing rates and alert someone if breathing stops without any tubes or wires connected to the patient.
In addition to other possible uses, the study authors want to conduct research with doctors to test the method as an infant-breathing monitor, and, if it proves useful, develop it as a medical device that would need federal approval. It may be useful for adults with sleep apnea.
Because of efforts to patent the new use of the wireless breathing-detection technology – which has been named BreathTaking – Neal Patwari posted his study on the online scientific preprint website ArXiv before submitting it to a journal for formal publication.
Patwari conducted the study with Wilson; Sai Ananthanarayanan, a postdoctoral electrical engineer; Sneha Kasera, an associate professor of computer science; and Dwayne Westenskow, a professor of anesthesiology and research professor of bioengineering. The research was funded by the National Science Foundation.
In a new study, Patwari showed a network of 20 wireless transceivers placed around a hospital bed could reliably detect breathing and estimate breathing rate to within two-fifths of a breath per minute based on 30 seconds of data.
This is different than using wireless transmitters to relay measurements from conventional breathing monitors. The motion of the chest and abdomen during breathing impedes the wireless radio signals crisscrossing a bedridden patient, who in the study was Patwari himself. Each of the 20 transceivers or "nodes" can transmit and receive to the other 19, meaning there can be up to 380 measurements (20 times 19) of radio signal strength within a short period of time (the transceivers transmit one after the other).
The study was conducted in a clinical room used for research at the University of Utah School of Medicine's Department of Anesthesiology. Patwari reclined on a hospital bed and listened to a metronome to time his breathing so he inhaled and exhaled 15 times per minute – about the average breathing rate for a resting adult.
His breathing was measured two ways: by the experimental wireless network, and by a carbon dioxide monitor connected to his nostrils by tubes. It calculated breathing rate by measuring the amount of carbon dioxide exhaled with each breath. Patwari also tested the wireless network with no one in the hospital bed.
The study found the wireless network could measure breathing within 0.4 to 0.2 breaths per minute, an insignificant error rate given that most breathing monitors round to the nearest breath per minute, he says. If a bedridden person or baby moves, the wireless system detects the movement but cannot measure their breathing at the same time.
To decide if someone is breathing or not, the wireless system uses a computer algorithm – basically, a set of formulas. Patwari says his algorithm squares the amplitude or loudness of the signal on each link between nodes, then averages it over all 380 links. A number larger than 1.5 indicates breathing has been detected.
Patwari also measured how many nodes were required to measure breathing accurately. The minimum was 13 nodes or transceivers, while the rate of incorrect breathing measurements fell to zero when 19 nodes were used. The study also showed the height of the nodes around the hospital bed didn't significantly affect breathing measurements.
Patwari plans more research on whether different or multiple radio frequencies might detect breathing better than the one 2.4 gigahertz frequency used in the study.
He also wants to test whether the system can detect two people breathing at the same rate but not in sync – something that might make it possible to design a system that could detect not only the location of hostages in a building, but the number held together.