Robotics for Gait Training After Spinal Cord Injury Jose A. Galvez and David J. Reinkensmeyer Downloaded from http://meridian.allenpress.com/tscir/article-pdf/11/2/18/1982806/damj-g43a-16eh-1bdk.pdf by guest on 25 September 2021 Locomotor training can help people with spinal cord injury to regain some mobility. However, the amount of hands-on therapy that patients can receive is limited, as economic pressures are inherent in the health care system. Therefore, worldwide efforts are being made to automate locomotor training. Automation has the potential to make therapy more affordable and thus more available for more patients and for more time. This article reviews efforts to automate locomotor training. We suggest that developing robotic devices that assist only as needed during step training is an important direction for future research. Key words: biomedical technology, gait, locomotion, lower extremity, rehabilitation, robotics, therapy obots are devices that can move and vous system diseases, including traumatic exert mechanical forces in a con- brain injury, multiple sclerosis, Parkinson’s Rtrolled way. They are usually de- disease, and spinal cord injury (SCI).2 signed to substitute for human labor in tasks Studies of movement therapy after stroke that are dangerous or difficult for humans to and SCI have shown that recovery is better perform. Robotic capabilities differ from when more therapy is given.6–9 More therapy those of humans; robots excel at accuracy, can be given if robots are built that can repeatability, and speed but are typically not replicate the exercise provided by rehabilita- as adaptable or intelligent as humans in the tion therapists. Robots do not get tired, and ways they perform their tasks. they have the potential of delivering training Robots have been studied and used as an more consistently and of quantifying and assistive technology, such as an aid to daily controlling the variables associated with the living of impaired people, since the 1960s (see, e.g., Hillman1 for a review). Their use for therapy, such as facilitating exercise and Jose A. Galvez, PhD, is Postdoctoral Researcher, Department of Mechanical and Aerospace Engineer- subsequent recovery of motor function, is ing, University of California, Irvine. more recent, growing during the 1990s (see, e.g., Reinkensmeyer2 for a review). A major David J. Reinkensmeyer, PhD, is Associate Professor, Department of Mechanical and Aerospace Engineer- research focus has been the development of ing and Department of Biomedical Engineering, Uni- devices that can provide movement therapy versity of California, Irvine. for the arm after stroke.3–5 It may also be possible to use therapeutic robotics to treat Top Spinal Cord Inj Rehabil 2005;11(2):18–33 © 2005 Thomas Land Publishers, Inc. impairments resulting from other central ner- www.thomasland.com 18 Robotics for Gait Training 19 Downloaded from http://meridian.allenpress.com/tscir/article-pdf/11/2/18/1982806/damj-g43a-16eh-1bdk.pdf by guest on 25 September 2021 Figure 1. Locomotor training with body-weight support on a treadmill. assistance more accurately than human partial body-weight support.16,17 therapists. Robots also have the potential of Partial body-weight support treadmill implementing dynamic regimes of training training (PBWSTT) is a rehabilitation tech- that are beyond the capacities of human nique based on the animal studies of spinal therapists. cord plasticity. Three therapists assist the Increasing scientific evidence has accu- legs and hips of the patient walking on a mulated that training improves recovery of treadmill while part of the patient’s body mobility in patients with SCI. Studies with weight is supported by an overhead harness cats decades ago demonstrated that the mam- (Figure 1). PBWSTT is based on the prin- malian spinal cord can learn after injury.10,11 ciple of generating normative, locomotor- Cats and rodents with complete spinal cord like sensory input to promote the functional transection regain hindlimb stepping ability reorganization and recovery of the injured after training with partial support of the body spinal cord circuitry.18 In automated weight and assistance of leg movements on a PBWSTT, the three therapists shown in Fig- treadmill.10–13 In humans, patients with both ure 1 are replaced by a robotic system; thus acute and chronic incomplete SCIs improve only one therapist is required for initiating functional mobility both after a period of and monitoring the activity. Automated conventional mobility training tech- PBWSTT is arguably less strenuous for the niques14,15 and after treadmill training with therapist than any other manual locomotor 20 TOPICS IN SPINAL CORD INJURY REHABILITATION/FALL 2005 training technique. Automation has the po- tients, 92% of those initially wheelchair- tential to make locomotor training available bound became independent walkers after to more people, more intensively, for more PBWSTT, in contrast to only 50% after time and also can provide training in chronic conventional therapy. Among the chronic stages of SCI and even months or years after patients, 76% of those initially wheelchair- injury. bound learned to walk independently after Downloaded from http://meridian.allenpress.com/tscir/article-pdf/11/2/18/1982806/damj-g43a-16eh-1bdk.pdf by guest on 25 September 2021 PBWSTT compared to only 1/14 after con- Clinical Studies of Manually Assisted ventional training. Locomotor Training In a follow-up study,22 it was found that this functional recovery was maintained or The first studies that reported improve- improved in most patients 6 months to 6 ments in ambulatory capacity in chronic in- years after locomotor training without fur- complete SCI patients who underwent ther training. A prerequisite to maintaining PBWSTT date back to 1989–199216,19,20 from these long-term beneficial effects appears to two independently working groups. be a minimum of regular activity, such as In 1994, a third research group21 included walking daily at least short distances.23 This acute complete paraplegic patients in their is consistent with observations made in study study and showed that coordinated stepping of cats after SCI that indicated that hindlimb movements could be induced on the tread- stepping ability on a treadmill was main- mill in complete SCI patients. Leg extensor tained 6 weeks after cessation of training but electromyographic (EMG) activity in- decreased significantly after 12 weeks.24 creased during training, while the inappro- When subsequently retrained, the spinal cats priate, non–locomotor-like EMG activity regained stepping ability more rapidly than decreased. However, the group of complete when trained initially. SCI patients did not improve their ability to More recently, a large multicenter ran- walk over ground; their improvement was domized clinical trial with acute incomplete limited to stepping on the treadmill with SCI patients compared PBWSTT at high partial body-weight support. Patients with treadmill speeds with conventional mobility acute incomplete paraplegia improved their training.25 Preliminary reports indicate that overground walking. both groups improved their outcome mea- In 1995, Wernig and colleagues17 re- sures relevant to walking performance, but ported the first study that included a com- no significant differences were found be- parison with a control group that underwent tween the PBWSTT and the conventionally conventional training. Eighty-nine patients trained groups.26,27 Interventions started from with incomplete SCIs (44 chronic and 45 2 to 8 weeks after onset of injury; the total acute) underwent daily PBWSTT and were number of patients was 140. The convention- compared with 64 patients (24 chronic and ally trained group received an equal duration 40 acute) treated conventionally. The of therapy consisting of standing and step- therapy duration varied between patients in ping using a tilt table, standing frame, paral- a range from 1 to 5 months. The PBWSTT lel bars, or over ground. An article reporting group improved their mobility more than the results has not been published yet, and the control group. Among the acute pa- thus it is difficult to determine why the results Robotics for Gait Training 21 of this study may differ from those of Wernig two footplates that attach to the patient’s and colleagues.17 feet. They are driven by a singly actuated Even if PBWSTT is only as effective as mechanism that moves the foot along a conventional training, it is of particular inter- fixed ellipsoid-like trajectory with a est because one can argue that it is more doubled crank and rocker system.28 The amenable to automation than conventional sagittal movements of the torso are driven in Downloaded from http://meridian.allenpress.com/tscir/article-pdf/11/2/18/1982806/damj-g43a-16eh-1bdk.pdf by guest on 25 September 2021 locomotor training techniques. PBWSTT is a phase-dependent manner by ropes at- done on a stationary set-up with the patient tached to the harness and connected by held in a fixed reference position and orienta- another crank to the foot crank. The body tion. This makes PBWSTT easier to equip weight is unloaded as needed by an over- with the appropriate tools and robotic systems head harness. The GT I is installed in tens of than conventional training techniques, be- locations, mainly across Europe. Clinical cause the robotic devices do not have to move trials have been carried out with the GT I, themselves over ground along with the pa- and a more advanced version has been de- tient. Conventional training techniques signed and built, as described below. change when the patient progresses, from tilt table or standing frame to parallel bars and Lokomat overground walking, which adds to the com- The Lokomat (Hocoma Medical Engi- plexity of its automation. An automated neering, Inc., Zurich, Switzerland; Figure PBWSTT system addresses the problem of 2B) is a motorized exoskeleton worn by training (both traditional or PBWSTT) being patients during treadmill walking.29 Four an arduous and labor-intensive job, particu- rotary joints move hip and knee flexion/ larly with severely impaired patients.