Stroke is one of the leading causes of disability in the United States. After a stroke, patients require both rehabilitation to reduce upper and/or lower body impairments, and assistance to help carry out daily living. The amount of conventional therapy performed is often limited due to cost and lack of intensity, therefore patients do not receive an effective dose to maximize gains.
Robotic training devices can promote greater dosage of therapy, but often are expensive and bulky, limiting their use to the clinic. In the home, wheelchairs provide low-cost mobility, but no means of therapy. Emerging hybrid assistive/rehabilitation devices like exoskeletons provide mobility with some clinical benefit but remain infeasible due to complexity and cost.
The premise of this dissertation is that simple, low-cost devices can be designed to promote both therapy and mobility, improving patient outcomes. This dissertation explored the design and experimental validation of three such devices, including a new class of hybrid assistive/rehabilitative device called an “exochair”. An exochair couples a rehabilitative exoskeleton to a wheelchair, with the goal of allowing a user’s therapeutic exercise drive around the chair, granting the user both assistance in daily living as well as therapy.
First, I report results from experiments that quantified the ability of young nonimpaired users to learn to drive a previously developed upper extremity exochair called LARA. The LARA chair is driven via levers with an arm support (i.e. by a simple exoskeleton) and is intended to help individuals with weak arms to self-propel and exercise the arms. I compared two designs of LARA, one that used a simple one-way bearing mechanism to provide propulsion (a technique that had been previously proposed), and a second that utilized a yoked hand-clutching scheme for driving (a novel technique suitable for people with unilateral arm/hand weakness such as occurs after stroke, but potentially more difficult to learn). Unexpectedly, I found that the participants learned to drive the hand-clutched device with similar learning time constants, and ultimately achieved similar over ground speeds, compared to the one-way bearing model. The hand-clutched approach increased the physiological effect of exercise, cognitive engagement, and maneuverability, showing its potential as a hybrid assistive/rehabilitation device for people with stroke.
In a follow-up study, I then quantified the ability of people with severe arm impairment after chronic stroke to learn to drive the hand clutch design. All participants increased their propulsion speed across five days of training. Unexpectedly, they showed a rate of motor learning that was two times higher than the average value reported for learning of a variety of motor tasks by unimpaired individuals, including the rate of learning of LARA by unimpaired users. They also improved on clinical arm movement scores by an amount comparable to previous robotic therapy device studies and showed high levels of motivation and self-ratings of competence that increased to high levels across training. Thus, the exochair unmasked a latent motor learning ability for individuals who have a severe arm impairment after stroke. This is significant because such arm impairment normally prevents such individuals from learning functional tasks with the impaired arm. People can learn to drive an exochair, it is rewarding, it assists in functional ambulation ability, and this learning process improves clinical arm movement scores.
Based on these experiments, I explored the application of classical machine learning algorithms to assist people with bilateral hand impairment in driving LARA. I showed the feasibility of an automated clutch system, comparing it with a hand switch approach. This approach could potentially relieve the burden of hand-clutching when using LARA for longer periods of time.
Next, I present results of a low-cost hand exoskeleton designed for assisting in hand extension post-stroke, with a demonstration of use by an unimpaired subject. This work explored an alternate way functionality could be increased for patients via wearable devices.
Finally, I extended the exochair concept to a hybrid assistive/rehabilitative device for leg movement called GRAM. My design concept for GRAM was that users could manipulate a linkage with their legs, mimicking the natural kinematic pattern of walking, in order to propel the wheelchair. I tested GRAM in a case study with an unimpaired individual, demonstrating that it was possible to learn to propel GRAM using this technique.
These results demonstrate the potential of exochairs such as LARA and GRAM to provide hybrid assistance/therapy in a simpler and less costly way than is currently possible with other technologies. Future research will study the long-term functional and therapeutic benefits of exochairs in larger patient population.
