Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

Wireless Sensor System Detects Occurrence of Elderly Falls

Fall detection radio-frequency sensor

Fall detection radio-frequency sensor (Dan Hixson, University of Utah)

Engineers at University of Utah in Salt Lake City developed a system combining wireless radio-wave sensors and a control algorithm to detect a person falling, without the individual wearing a separate device. Graduate student Brad Mager, representing the Utah team, presents the findings of a proof-of-concept test of the system today at the IEEE Personal, Indoor and Mobile Radio Communications Conference in London.

Mager and senior author Neal Patwari, an engineering professor at Utah, plan to commercialize the technology through Patwari’s three year-old company Xandem Technology, also in Salt Lake City.

In their paper, Mager and colleagues cite studies showing that for people age 65 and older living in their own homes, 30 percent fall each year, with about half of older people in residential living facilities falling each year. As a result, falls are the leading cause of injury for people age 65 and older, who are expected to number 1.5 billion worldwide by the year 2050. While preventing falls is the most desirable solution for this problem, the next best thing is to detect falls as soon as possible after they happen, and get help to the person who fell.

Systems are available today that are worn by individuals and designed to either detect a fall or provide a button to press that alerts caregivers. These systems require the person to wear the detection or call device all the time, even at night, which may not be realistic. Also, a 2008 study in British Journal of Medicine (cited in the conference paper) reports that the vast majority of people over the age of 90 with these devices who fall do not have the call device on them when they fall, putting it out of reach and thus unusable in that emergency.

Technologies are available to monitor older people in their homes, without wearing a separate device. Some of those technologies however, such as live video, are considered too invasive. Other technologies, such as passive infrared sensors, can sense a lack of motion and infer that a fall has occurred, but while less invasive than live video, they also run the risk of false alarms if, for example, a person gets up at night, moves to another room, and falls asleep there.

The Utah team’s solution offers a technology called radio tomography that places radio-frequency (RF) transceivers at various locations and heights in a room to sense an individual’s three-dimensional presence as well as orientation: standing, sitting, or prone (lying down). The presence of a human body, composed mainly of water, causes radio waves  to alter their propogation as people move around the room. The radio signals can detect the presence of objects as small as six inches, and are not blocked by materials other than metal, thus the RF devices can be hidden behind walls or inside objects, such as vases.

While the sensors at multiple levels and locations can return abstract images representing the presence and orientations of individuals, detecting falls requires adding to the system the ability to discriminate between normal changes in position and abrupt movements indicating an abnormal condition. The Utah system processes the measurements of individuals moving around the room with an algorithm based on a hidden Markov model, a statistical technique that allows for making intuitive judgements if the true state cannot be directly observed. If the signals indicate a person’s orientation changes abruptly from standing to prone, rather than deliberately sitting or lying down for example, the algorithm would detect that abrupt change as a fall.

Mager, Patwari, and colleagues tested the system in a simulated living room with 24 RF sensor devices placed at two levels: 17 (6.7 inches) and 140 centimeters (55.1 inches) from the floor. The researchers ran a series of 54 tests, each 40 to 55 seconds in length, where volunteer subjects performed various pre-arranged motions, including simulated falls on a padded floor, to allow for realistic trips and crumbles. The results showed the system with its combination of sensors and algorithm discriminated between normal movements and falls with 100 percent accuracy.

The Utah team plans further tests with a wider variety of body types, more cluttered environments, and longer periods of time to simulate a wider range of human motions.

Read more:

*     *     *

1 comment to Wireless Sensor System Detects Occurrence of Elderly Falls