Stroke victims face a difficult and expensive road to recovery that is mentally and physically straining, but there may be a new answer according to one University of Massachusetts researcher: robots.
Because therapists are limited and expensive, and stroke rehabilitation requires extensive and time-consuming therapy for patients to regain their ability to communicate and control their muscles, researchers are looking to robotics for a cheaper solution.
University of Massachusetts assistant professor of communication disorders Yu-Kyong Choe was recently been awarded a $109,251 grant from the American Heart Association (AHA) to study the use of humanoid robots in facilitating stroke rehabilitation.
Choe, a speech therapist, is collaborating with Roderic Grupen, director of the Laboratory for Perceptual Robotics to conduct the study. Choe and Grupen will be assisted by graduate students studying communication disorders and computer science.
A stroke occurs when blood flow to the brain is obstructed, preventing oxygen and glucose from getting to the brain and damaging the tissue.
“If you have a stroke that affects the language center in the left brain, you will have difficulty moving your right arm and retrieving important words,” Choe said.
Patients can often work out what they’d like to say in their mind, but they have trouble connecting their ideas to words and sounds.
She has developed software that patients can use on their home computer to practice word retrieval tasks. The program works by displaying pictures on the screen and providing cues to help the patient recall the correct words to express themselves.
“Today they may need 5 or 6 cues, but tomorrow they’ll need 4,” Choe said. “The process continues until they can independently recall the words on their own.”
The purpose of the AHA study is to load Choe’s software onto a child-sized robot that can also help patients with their physical therapy. They will explore how this compares to software-based therapy delivered from a laptop computer and how patients and therapists respond to each method.
“When this robot presents a physical target, you really have to go there,” Choe said.
However, she can’t do the physical part on her own.
“I’m a speech therapist so I can do the speech part, but I need to have an occupational or physical therapist who can assess, treat and develop arm exercises,” she said.
The challenge is developing a robot that can live with patients at home and learn from its interactions with them. Roderic Grupen and his team look to how animals play and explore in their environment and try to replicate that behavior in robots.
Animals utilize play as a fundamental survival strategy, to “find out what [their] relationship to the world is,” he said.
Animals will also behave in a certain way if their actions bring about a reward, such as giving a dog a treat for sitting down They learn to associate sitting with a prize and warm praise, so eventually they’ll sit on command.
Robots can learn in a similar way. The dog treat equivalent for a robot is a “surprising event,” or what’s known as an “intrinsic reward,” Grupen said. A robot, equipped with audio and visual sensors, may see a patient on its left side. When the patient moves from that spot, the robot will register that unusual movement and try to make sense of it.
“This is an ambitious machine learning project driven by intrinsic motivation,” Grupen said.
In this way, the robot can learn about the special needs of the client it’s working with, such as the layout of the house. The robot would also log the progress of the patient and that information would be available to the therapist.
Choe and Grupen acknowledge that some people may think of robots are replacing human interaction, but they see the robots as an extension of the therapist’s toolkit. Grupen called the technology a way to “attack isolation” and foster human-to-human interaction.
The robot can provide many hours of intensive therapies in between live sessions that are crucial for functional recovery from a stroke.
Choe points to the recovery of Congresswoman Gabby Giffords as an example of how intense therapy can lead to successful rehabilitation. She said if other patients had access to similar therapy, they would also see great strides in recovery.
“It doesn’t really matter if a person had a stroke yesterday or 10 years ago. Once we bring them into our clinic they start making progress,” Choe said.
To participate in the study or learn more about the project, contact Choe at (413) 545-4297 or send her an e-mail to: [email protected].
David Barnstone can be reached at [email protected].
