Take one robot, add a simple video game and in the hands of the right researchers you get a novel new device that helps stroke patients recover more quickly. An initial clinical trial showed that patients using the robot-assisted therapy improved faster than those who did not use the robot.

"What I want to do is get robotics and information technology into clinical neurology," explained Professor Neville Hogan of the Massachusetts Institute of Technology, principal investigator for the project.

He is well qualified to carry out such work given that he has a dual MIT appointment in two departments, Mechanical Engineering and Brain and Cognitive Science. He also happens to be from Dublin and did his primary degree at Bolton Street College of Technology.

"I have been interested in rehabilitation systems for some time," he said. He did his PhD on the engineering of artificial limbs and has since studied these and robotics as it relates to the control of movement. Robots, he points out, have very little subtlety of movement and can easily be "destabilised". It is difficult to achieve machine dexterity, he adds. "You don't see many robots that can juggle."

His dual chair allows him to integrate robotics and engineering with biology and movement control. In particular he is looking at how robots could help neurological rehabilitation after stroke, which he said, is the "leading cause of motor disability in the US".

Rehabilitation after stroke involves developing "structured physical activity" for the patient's limbs, he said, and the new device automates this activity. His new robot is named MITManus which links its therapeutic focus and MIT's motto "mens et manus" - mind and hand, particularly apt given its unusual design. The patient sits at a table and puts the lower arm and wrist into a brace that is in turn attached to the "arm" of the robot.

The therapist can move the patient's hand through a sequence of exercises, with MIT-Manus recording them in detail. It can then reproduce these movements, manipulating the patient's arm automatically.

It provides much more however. It can guide the patient through the routines but can also respond to the patient's own movements once the person begins to recover. MIT-Manus can measure how much force the patient is applying and adjust the amount of resistance it applies.

The exercises are also linked to a simple video game with the patient attempting to move a cursor towards a target. "We wanted to give the patient visually evoked movement. They do that by interacting with the handle of the robot," he explained. "The machine is interactive."

The game's level of difficulty can be adjusted so that the target can be hit between 70 and 90 per cent of the time no matter what the patient's motor ability. If there is very little motor control the robot moves the patient, but when control returns then the patient moves the robot. "The patient always succeeds to some degree," and this he said provided motivation to continue.

The robot has been used in an initial clinical trial involving 20 stroke patients. All received standard therapy but half were given four or five hours a week on MITManus and half were given "sham" robot therapy once a week. Those using the robot improved further and faster than the control group, Professor Hogan said.

The sample was small and it has yet to be shown whether it was the robot or the additional exercises that led to the improvement, but the trial did show that the robot technique could be used and was satisfactory for such patients. It also proved that manual manipulation of a stroke victim's disabled limb does aid recovery. "There had been a great deal of intuitive belief that this works, but until now there was no objective evidence," he said.

A new trial involving 60 patients is already under way and would be completed by August, he said.

"We think that this will have an impact on the type of therapy you do for these patients," Professor Hogan added.

His colleagues included Dr Hernano I Krebs, a research scientist at MIT and Dr Mindy L Aisen, Dr Fletcher H McDowell and Dr Bruce T Volpe of the Burke Rehabilitation Centre in White Plains, New York, where the trials are taking place.

Professor Hogan continues to carry out research on the robot and is establishing a company to develop it as a commercial product. He pictures a team of MIT-Manus robots working with patients and all under the control of a single therapist who, like a chess grand master playing 10 games at once, monitors the progress of each of the patients in great detail.

It is also a useful new research tool in cognitive studies which explore how the brain works and how limb movement is controlled.