Each year, two million Americans suffer brain injuries or strokes that can impair their ability to move their limbs. Traditional physical therapy can help patients compensate for the damage, but many patients tend to reach a plateau in performance after several weeks.

Since the early 1990s, however, a few patients have been able to continue their progress thanks to an experimental robot built for arm rehabilitation. It never tires, adjusts as the patients improve, and precisely measures and monitors their performance.

Now, that machine – plus three similar ones – are moving into large-scale tests equivalent to late-stage drug trials, the first such trials for therapeutic robots.

Run by the Veterans Health Administration, the new clinical study will involve approximately 200 patients. The randomized trials, set to begin next year and run for three years in an as-yet-undetermined number of hospitals, will test the robots head-to-head with traditional therapy.

If all goes well, it “could provide the evidence needed to adopt the robot’s clinical use throughout the VA system and possibly beyond,” says Albert Lo, a Yale University neurologist and principal investigator in the trials.

The robots were built by MIT mechanical engineers Neville Hogan and Hermano Igo Krebs, who founded Interactive Motion Technologies of Cambridge, MA, to commercialize their work. The pair’s first and most extensively tested device is a two-jointed motorized arm that glides parallel to a desktop. Patients grasp a handle and attempt to move it in and out, and left and right, giving their shoulders and elbows a workout.

Unlike physical-therapy treatments that simply move a patient’s limbs repeatedly in a pattern, the arm robot enlists the patient’s participation in the therapy, providing help only when needed.

The robot’s software adjusts to the patient’s progress. People starting out may not even be able to move their arms; at this stage, the robot fully impels and guides their movements. As patients improve, the robot gradually reduces the assistance it provides. At some stage, it may no longer help them move, instead only guiding movements along certain paths. Or it might wait before lending a hand, giving patients more of a chance to perform movements on their own.

The device also has a video game component that directs the therapy and helps keep the patients motivated. The patients’ movements guide a cursor on a computer screen toward targets. The computer regularly shows patients how well they are doing.

Stanley Schaffer, of Scarsdale, NY, who enjoyed playing classical piano until a stroke paralyzed one of his arms, says seeing his progress inspires him to keep trying. “You’re competing against yourself,” says Schaffer, who in a smaller-scale trial this year used the arm robot and a similar one for the wrist. With traditional therapy, he had stopped seeing improvements. But the robots have helped improve his mobility. “I feel that this is definitely going to help me get back to playing my piano with two hands,” he says.

Interactive Motion’s exercising robots are not alone. Companies working to commercialize therapeutic robots that focus on walking and balance include Chicago PT of Evanston, IL; Hocoma of Volketswil, Switzerland; Robomedica of Irvine, CA; and Yaskawa Electric of Tokyo, Japan. Taken together, corporate efforts and the latest clinical trials mean robots could eventually find widespread use in physical therapy.