MIT engineers have developed a robot that can "learn" exercises from a physical therapist, guide a patient through them, and-for the first time-record biomechanical data on the patient's condition and progress.
As a result, the robot could not only assist a busy therapist by performing repetitive exercises but also quantify forces and movements that until now have been judged by "touch and feel." Such data would provide a permanent, objective record of a patient's progress that could be used in many ways.
"We're going to put an aspect of physical therapy onto the information highway," said Neville Hogan, principal investigator for the work and an MIT professor with dual appointments in mechanical engineering and brain and cognitive sciences.
This fall the robot, which specifically aids therapists in exercises for the wrist and hand, will be tested in clinical trials with people who have suffered strokes.
The novel ability to record a patient's biomechanical performance "could have applications for a number of other disorders, but we wanted to target the largest group that this machine could help," said Hermano I. Krebs, a graduate student in mechanical engineering who is developing the robot for his doctoral thesis. Some 250,000 Americans have strokes every year, of whom "about 60 percent could potentially benefit from this machine," he said.
Professor Hogan and Mr. Krebs will use the trials, which will be held at the Burke Rehabilitation Center in White Plains, NY, to see how the machine works in a clinical setting. In addition, the MIT engineers and colleagues at Burke hope to answer a long-standing question among neurologists. Does manual manipulation of a stroke victim's limbs really help that person recover?
"At present, although there is a great deal of intuitive belief that this works, there is no objective evidence that manual manipulation of a stroke victim's limbs makes any difference at all," said Professor Hogan, who is also director of MIT's Eric P. and Evelyn E. Newman Laboratory for Biomechanics and Human Rehabilitation. The robot-named Manus for the link between its therapeutic focus and MIT's motto "mens et manus," or mind and hand-could supply the objective data to answer that question. It is capable of recording the amount of force a patient applies, the velocity of movements, the position of the hand and more.
In a three-month experiment, two groups of stroke patients will each receive standard physical therapy. People in one of the groups will receive additional manipulations provided by Manus. Just before the experiment starts and at two-week intervals thereafter, the researchers will use the robot to collect data on participants in both groups. "Then we'll see if we can detect a difference in functional outcome between them," Professor Hogan said.
Neurologists at Burke who are collaborating with the MIT engineers are Drs. Fletcher H. McDowell, Mindy Aisen, Bruce T. Volpe and Michael Reding.
To use Manus, a person sitting at a table puts the lower arm and wrist into a brace that is in turn attached to the "arm" of the robot (Professor Hogan noted that patients can pull out of the brace completely in an emergency). With "hand over hand" instruction, a therapist then physically guides the patient through a given exercise. The machine, which allows movements in three degrees of freedom, "goes along for the ride" and records the session, Professor Hogan said.
Manus can then reproduce the exercise and guide the patient through it. And as the patient begins to recover and starts to initiate some movement on his or her own, the robot can measure how much force the patient is applying and adjust the amount of resistance it provides.
The engineers have also thought of a way to hold a patient's attention during what can be monotonous exercises: Manus is linked to video games. "It's one of the things our neurologist collaborators got most excited about," Professor Hogan said.
Specifically, the games correspond to a patient's performance of a given exercise. "They provide a visual clue that a patient is, say, not pushing hard enough, or pushing too hard," Mr. Krebs said. (The computer screen on which the games appear will sit on the table in front of the patient.)
The researchers have developed four games so far to illustrate the concept, but the graphic possibilities "are up to your imagination," Professor Hogan said. "There's an enormous range of educational video games on the market, and we could certainly plug into that."
Mr. Krebs noted that the games are designed for patients at different levels of recovery. Even for people who cannot yet initiate any movement on their own, a visual display that includes such elements as a "happy face" symbol at the end of an exercise gives positive feedback. The researchers are also working to integrate music in the system.
Three undergraduates in mechanical engineering are working with Mr. Krebs on this part of the project. They are seniors Christopher M. Minekime and Amy S. Gorin, and junior Donna Louise Scott.
The MIT researchers envision a range of applications for Manus in addition to its use as a mechanical therapist.
For example, the device could create a permanent biomechanical record of the effects of a given drug on a patient's progress. The robot could also help doctors prescribe the optimum therapy for a new patient. By comparing data from the new patient with the records of previous patients, "they can get a better sense of where the new patient stands," Professor Hogan said.
Manus could also become an important teaching tool. "You could record how a skilled neurologist performs a given procedure on a mannequin, then replay it for trainees," Professor Hogan said. Themachine could repeat the exercise, for example, while a trainee physically follows the movements with his hands over the mannequin's wrist and hand.
And if two machines were linked, the expert could put an arm into one while the trainee works on a mannequin at the other. The data recorded by the student's machine would be communicated to the expert's hand, so the expert could "feel and observe what the trainee is doing," Professor Hogan said.
The data recorded by the robot could also be sent over the information highway linking, say, a therapist in a small town with a specialist in Boston. With robots in both offices, the two professionals could work together to solve a particularly perplexing problem by transmitting the actual "touch and feel" of the patient.
Manus grew out of years of research in Professor Hogan's lab on the problem of robotic interactions with humans. "Most robots today operate in a very structured environment," he explained, because if unexpected variables are introduced-like bumping into a human being-the machine can "oscillate uncontrollably, with potentially devastating results."
"We came up with one solution to the problem, so the next step was to develop an interesting application," Professor Hogan said. Manus is the result.
The MIT engineers have applied for a patent on Manus. The work has been supported by the National Science Foundation and the National Institute on Disability and Rehabilitation Research.
A version of this article appeared in the March 16, 1994 issue of MIT Tech Talk (Volume 38, Number 26).