The same University of Utah engineers who built wireless networks that see through walls are now redirecting their efforts to noninvasively measure the breath of hospital patients and newborns.
Elizabeth Armstrong Moore
Elizabeth Armstrong Moore is based in Portland, Oregon, and has written for Wired, The Christian Science Monitor, and public radio. Her semi-obscure hobbies include climbing, billiards, board games that take up a lot of space, and piano.
Engineers at the University of Utah predict that, in about five years, a network of wireless transceivers around a bed will be able to measure breathing rates without a single tube or wire being connected to the patient.
The uses of the system, which the team has dubbed BreathTaking, are obvious: patients in post-op, infants at risk of Sudden Infant Death Syndrome, or even people with sleep apnea.
And because the technology uses off-the-shelf transceivers similar to ones used in home computer networks, the system could cost less than current breathing monitors, said electrical engineer Neal Patwari, senior author of the team's study, which he posted to the online scientific preprint site ArXiv before submitting it for publication in a scientific journal.
In 2009, Patwari and then-grad student Joey Wilson demonstrated that a couple dozen wireless transceivers could be used to see through walls for the detection of, say, a burglar, hostages, or even kids partying when the parents are away. (They are now commercializing these networks for use as next-gen motion detectors.)
For this work, Patwari found that a network of 20 wireless transceivers around a hospital bed was able to estimate breathing rate to within two-fifths of a breath per minute based on 30 seconds of data.
Because the chest and abdomen move during breathing, which impedes the crisscrossing signals, each of the nodes can receive and transmit to and from the other 19, resulting in as many as 380 measurements of radio signal strength.
To detect breath, a computer algorithm squares the amplitude of the signal on every link between nodes and averages that volume over the 380 links. Patwari also found that at least 13 nodes are required to detect breath, and that 19 was the minimum for near 100 percent accuracy.
Patwari says he plans to research whether different (or multiple) radio frequencies could further improve on the 2.4 gigahertz frequency he used in this initial study, as well as investigate the possibility of detecting more than one person's breathing at the same time for use in locating, say, hostages in a building.