The University of Massachusetts Amherst

Feature Stories

Proactive Health

Researchers test and validate wearable sensors to improve health management
  • Graduate students test the equipment on the treadmill in the lab.

Freedson notes that self-reporting has historically fallen short of providing accurate data, which is why these wearable sensing devices are changing the face of activity assessment. 

As health professionals find increasingly more evidence linking physical activity to health, new devices like the “Fitbit” and smartphone apps like “My Fitness Pal” have made personal activity monitoring markedly easier. These evolving technologies require the continued experimentation of UMass Amherst researchers like Patty Freedson, who works to map new methodologies for monitoring activity behavior.

Freedson, chair of the Kinesiology Department (School of Public Health and Health Sciences) and director of the Physical Activity and Health Laboratory, has studied the objective assessment of physical activity with wearable monitors for more than twenty years. Her research, funded by the National Institutes of Health (NIH), has grown along with the blossoming field. Accelerometers, devices that measure the amount and intensity of movement, have drastically dropped in cost and are now readily available in the smartphone—an advance that allows Freedson and her colleagues to have more impact than ever before on health outcomes.

Freedson and statistician John Staudenmayer (Mathematics and Statistics) have primarily worked with an accelerometer manufactured by ActiGraph, a company that produces accelerometer hardware and data management software.  Their research results are used by ActiGraph to translate accelerometer signals into various physical activity metrics. They and graduate students Sarah Kozey Keadle, Kate Lyden, Amanda Hickey, and Jeffer Sasaki have also worked with the activPAL sensor manufactured by PAL Technologies that is used to detect and monitor sitting behavior. Emerging evidence suggests that, independent of physical activity, sedentary behavior is associated with health.

“Very consistent positive relationships have been reported between physical activity and health. Now that we have objective tools to assess physical activity dose, we can be more accurate in describing how much physical activity is sufficient for health,” Freedson says.

In one NIH funded project, Freedson, Staudenmayer and Lyden have developed and validated machine-learning algorithms that estimate activity level from the accelerometer signals. Machine learning, a branch of artificial intelligence, uses computer-generated models to “learn” from collected data. In this case, Freedson and the team directly measure energy expenditure using a portable respiratory gas exchange device and identify activity type. They then feed these data, along with the accelerometer signal features, into a computer—information that enables the activity monitors to recognize different types of activity and estimate the corresponding energy expenditure. These accelerometer-based data processing methods accurately quantify the amount and intensity of physical activity so experts can better understand the “dose” of activity that correlates to positive health outcomes.

In another NIH project, Freedson and her team are testing and validating sensors that estimate breathing volume. Besides monitoring and assessing physical activity, Freedson says there is interest in the devices for other applications, such as environmental exposure studies. Air pollution is an area of concern, particularly in urban areas, and such a device could allow environmental health researchers to quantify internal exposure to air pollutants. 

Freedson, Hickey, and Sasaki have also joined forces with physicians David Ayers and Patricia Franklin at the UMass Medical School (Orthopedics and Physical Rehabilitation) to study physical activity and sedentary behavior in patients with osteoarthritis. Patients in the study wear an accelerometer sensor that differentiates postural positions. With funding through the UMass Center for Clinical and Translational Science Moment Fund, Freedson quantifies how much sitting, standing and stepping osteoarthritis patients do and how these behaviors change during disease progression.

Freedson explains that simple self-reporting has historically fallen short of providing accurate data, which is why these wearable sensing devices are changing the face of activity assessment. As the technology continues to evolve, sensor design is yielding smaller devices with extensive memory capacity.

“People really like them because it provides an opportunity to self manage their physical activity behavior…similar to when you get on a scale or take your blood pressure, you have a number that tells you how much activity or how much sitting you do,” Freedson says.

Freedson’s work is an essential element of the campus’s Personalized Health Monitoring (PHM) efforts, focused on developing nanotechnology and large dataset management to improve health care through low-cost, wearable, wireless sensors that analyze patient data continuously in real time. Thanks to the research conducted by Freedson and her colleagues, both patients and exercisers will have new tools at their disposal to monitor physical activity and better control their health. 

Amanda Drane '12