Improving physiotherapy using sensors and machine learning

Tuesday, May 23, 2017

People recovering from hip and knee replacements perform rehabilitative exercises better when they get immediate visual feedback, initial testing of new technology has shown.

multi-year project at the University of Waterloo combined motion sensors with software programs to enable typically elderly patients to watch how they were doing on a computer screen compared to the target movement.

Researchers found the system, now being refined for possible commercialization, helped patients perform exercises with more control and less cheating. Some patients also increased their range of motion, a key objective of the physiotherapy.

Dana Kulić, a professor of electrical and computer engineering at Waterloo, said the findings suggest the system could help reduce recovery times by increasing the effectiveness of exercises, a hypothesis she hopes to test in a follow-up study.

More research is also required to confirm a link between visual feedback and better exercise performance.

In addition to joint replacements, the system has potential applications in rehabilitation for a range of other patients, including stroke victims and people with Parkinson’s disease.

“The system we have developed is quite flexible,” said Kulić, who presented a published research paper on the project at the recent Computer-Human Interaction Conference in Denver. “It can be used for any type of human movement.”

Sensors attached to limbs – just above the knee and the ankle in knee replacement cases, for example – send data to a computer while patients do exercises prescribed by physiotherapists.

Using human body modelling and machine learning, software then analyzes that data and, in addition to organizing and storing it, instantly generates a stick-figure visual representation of the motion.

The patient’s actual movement is displayed on a computer screen alongside the ideal movement for a given exercise, providing a comparison patients described as both motivating and enjoyable.

“Some even asked if it was possible to buy the system to take it home,” Kulić said.

A key function of the Automated Rehabilitation System is the provision of objective information on patient progress to physiotherapists, who currently rely largely on visual observation and experience to make assessments.

Cardon Rehabilitation and Medical Equipment Ltd. of Burlington, Ont., which partnered with Waterloo on the project, is now working to bring the technology to market.

The lead researcher was Waterloo master’s student Agnes Lam, who has since been hired by Cardon. Also involved were graduate students and researchers Jonathan Lin, Vlad Joukov, Roshanak Houmanfar, Michelle Karg, Danniel Varona-Marin and Yeti Li, engineering professor Catherine Burns and Mitchell Fergenbaum, a professor at Sheridan College in Oakville.

Testing was done at the Toronto Rehabilitation InstituteSt. Joseph’s Health Centre in Guelph and the Freeport site of Grand River Hospital in Kitchener.

  1. 2020 (1)
    1. October (1)
  2. 2019 (6)
    1. December (2)
    2. October (1)
    3. August (1)
    4. January (2)
  3. 2018 (10)
  4. 2017 (6)
  5. 2016 (2)
  6. 2015 (1)