Robotics in Rehabilitation Medicine: Prosthetics, Exoskeletons, All Else in Rehabilitation Medicine

Key Points

•

Robots assist with upper- and lower-extremity physical rehabilitation.
•

This technology can also provide therapies for social and emotional concerns.
•

Safety in design is a primary concern.
•

Robots can help meet the needs of a growing aging population.
•

The integration of robotics into prosthetic and orthotic devices follows a natural historic progression.
•

Patients can expect greater restoration of the function of lost limbs through robotic prostheses.
•

Improvements in these devices restore not only function but also a more natural appearance and motion.
•

Advances in the brain-computer interface and osseointegration promise encouraging new potentials.

Robotics has fascinated us since its conception. In popular culture, robots have been portrayed as helping with chores, as fighting alongside soldiers in battles, or as true friends with real emotional connections. It is clear that robots could assist humanity through integration into various different areas of our lives. Rehabilitation is one area in which robots could excel. A robot is potentially strong and durable and can automate tasks with exact reproducibility in situations where humans might otherwise become fatigued, leading to errors and variances. Robotic devices could capture and record data through sensors, then process and analyze those data and provide dynamic feedback to the end users. These features could be used for therapeutic, adaptive, and assistive purposes.

In rehabilitation, we can classify robots into two major categories: robots used for therapy, such as physical or occupational therapy, and robots used for assistance in daily living to increase independence and/or decrease caregiver burden. Robots designed to be used for therapy are used to improve an individual’s function by assisting in supervised professional therapy sessions such as strengthening and neurorestorative or cardiopulmonary conditioning therapies. Robots designed for assistance in daily living can support and augment a patient’s function or even replace their disabilities and limitations, ultimately to enhance a person’s independence and/or decrease dependence on others. However, robotics in rehabilitation is not limited to just these two broad categories. There are also robots that are designed to fulfill social interactions and emotional support.

Robotic rehabilitation therapies began in the early 1900s with a machine invented by Theodor Budigen to support stepping movements for patients with heart disease. By the 1930s, Richard Scherb had created early prototypical exoskeletons with cables and mechanical joints that aided in orthopedic therapy, mechanically helping to increase range and to stretch the patient’s limbs. Similarly, later industrial manipulator robots helped in therapies incorporating repetitive dynamic movements, where the robot would hold on to pads and move the pad according to the preplanned therapy regimen. Powered exoskeletons were invented around the 1970s utilizing pneumatic, hydraulic, and electromagnetic actuators for limb movement.

It was not until the 1980s that scientists first started designing robots for neurorehabilitation. In 1989, Dr. Krebs and his team at the Massachusetts Institute of Technology designed the robot MIT-Manus to allow finer control over the force and movements of the upper extremity. Lokomat, first introduced in 1994, is another pioneer robot in lower-body rehabilitation that supports body weight during treadmill training (see Fig. 6.1 ). In 2002, Reading University (UK) developed a robot that allows patients to train with virtual reality. Since then, there has been a proliferation of developments, many designed with neurorehabilitation in mind, that interact with patients through passive movements, as well as assisted-active and actively resisted movements, virtual reality, and sensory feedback (see Fig. 6.2 ).

Neurorehabilitation is a method in rehabilitation where neurorestorative therapies are employed with the goal of neurorecovery: the restoration of neurologic functions after a neurologic injury or insult. Neurorestorative therapies help patients who have lost sensorimotor control and function after neurologic injuries such as stroke, brain injury, and spinal cord injury. These therapies attempt to take advantage of neuroplasticity, based on the theory that the nervous system can be retrained. To reestablish neural pathways, therapy consists of sensory and motor stimulation through a passive range of motions, functional exercises, and strengthening programs. These types of therapy are often high intensity and repetitive, since a very high dose of these tasks is necessary to achieve meaningful gains. During a traditional therapy session, patients typically perform about 30 repetitions in a 45-minute session. However, robots are capable of consistently delivering a reproducible experience, and each session can consist of as much as 1000 repetitions. With the help of robots, therapists are able to offload highly repetitive and tiring tasks, and instead are able to focus more on finer nuances, overall therapeutic goals, and/or more efficiently seeing higher volumes of patients.

Robots are not just limited to repetitive tasks. Equipped with onboard sensors, robots can gather objective data, such as limb position, velocity, and force, to estimate muscle activation, as well as other physiologic states of the user such as cardiac output and respiratory effort. These data could be processed and analyzed in real time or stored in memory for clinicians to monitor progress and augment clinical decision-making to potentially enhance neurorecovery. Additionally, data can be analyzed through computational models for the robot to adapt to fluctuations in the patient’s performance and to make real-time changes to treatment. By providing direct feedback to patients and providers, robots enhance neurorecovery through optimal synchronization between patient motivation and their motor output.

One highly effective neurorestorative approach is called constraint-induced movement therapy. After one limb becomes less functional, the other limb compensates for the loss of function, thereby patients are unable to recover much function of the affected limb due to prolonged disuse. Constraint-induced movement therapy was developed to prevent learned disuse and maximize functionality of the affected limb. However, providing the right amount of assistance while maximizing patient effort to complete the tasks are difficult and depends on the severity of the impairment. Incorporating robotic therapies allow for fine-tuning and adapting as the patient fatigues or, conversely, recovers more strength.

In addition to motor impairments, sensory and cognitive impairments also contribute to decline in function. Sensory function is necessary in providing feedback to the brain, for example in maintaining appropriate strength while grasping an object or positioning the limb in space right before taking a step. Patients often rely on visual compensation; however, they also suffer from cognitive impairments, and these compensations could be limited.

Despite every effort for neurorecovery, not all patients recover completely. Even those who obtain partial function most rarely achieve over 70% to 80% of the previous baseline. To compensate for residual impairments, assistive robots can help maximize independence during activities of daily living and minimize caregiver burden. For example, some patients may benefit from exoskeletons that provide gravity support and movement assistance, allowing them to regain control of the surrounding environment.

This chapter will present basic concepts in designing robots for rehabilitation, how therapy robots are used for upper- and lower-limb recovery, as well as social and emotional domains and how assistive and other robots are contributing to rehabilitation.

Basic Concepts in Robotic Design for Rehabilitation

There are two major types of robotic design for rehabilitation, end-effector types and exoskeleton types. End-effector robots control and guide the most distal part of the patient’s body, such as a hand or foot, through the endpiece of the robot, and thus the whole system is mostly independent of the patient. Due to this independence, these robots have a greater degree of freedom, a larger field of space, and higher motion dynamics. In contrast, exoskeleton robots control one or more joints and are either integrated or attached directly to the patient’s limb. As exoskeleton robots are constrained by human joints, they have less freedom of movement but are more compact and better able to assist motion. Further discussion of exoskeleton robots will be presented later in the chapter.

The most important robotic design feature is safety (see Table 6.1 ). Often, excessive tensions can cause injuries to muscle, tendons, ligaments, and bones, therefore robotic devices should monitor and prevent any tissue damage while working within reasonable ranges of motion, speed, and force. These safety measures should be enforced through both mechanical constraints and software limiters. Furthermore, it is critical that all machines have prominent and easily accessible stop buttons. Despite all these precautions, patient safety ultimately lies with the machine operator who must set reasonable parameters for the robot and monitor and control the situation.

Table 6.1

Possible Safety Mechanisms

Safety Mechanism	Description
User initiated
Emergency button	Button provides an easy method for patient to override control unit for immediate stop.
Voice command	Robot listens for a “stop” voice command.
Unactuated digit detection	Patients control the unit through voluntary movement of digits that are not being actuated.
Sensors
Strain sensor	Sensor measures joint bending and stress on joint and shuts off if threshold strain is exceeded.
Pressure sensor	Sensor measures pressure in areas directly in contact with patient and shuts off if a threshold pressure is exceeded.
Mechanical limiters
Rotation limiter	Spools that guide cables are intrinsically limited in rotation to avoid hyperflexion or hyperextension of the limb.
Magnetic coupling	Cables are connected to magnetic ends and mechanically detach when tension exceeds the magnetic strength threshold.

Before the design phase, it is important to understand rehabilitation physiology and have a standardized therapy method that the robot could model. Without a standardized treatment plan, the robot and patient could begin repetitive motions in a slightly wrong position or trajectory, which could lead to learned mal-use and worse outcomes. In addition, robots must be able to recreate and reproduce intended motions according to the treatment plan while detecting for deviations and provide corrective measures or sound an alert.

The rehabilitation robot must be able to adapt to varying degrees of impairment in strength, range of motion, mobility, and coordination. Robots must be able to provide full assistance for a weak limb, including anticipating the patient’s intended motions or movements, which would be most helpful in adaptive robots assisting in daily living activities. However, robots must also be able to provide varying degrees of resistance for strength training. To achieve both these facets, the robotic actuators, or devices that convert control signals into mechanical motions, should have low impedance or resistance to change but a high capacity in delivering motion. Depending on specific therapy goals, the device should be able to change its focus between different parameters, such as range of motion, strength, speed, and coordination; be able to detect muscle spasms, while concomitantly adapting to varying inputs, patients, and environments; and be able to avoid collisions with other objects. The robotic system should be able to process the environment through its onboard sensors and adjust any limb malpositioning or trajectory deviation, and perhaps even individualize the treatment depending on the circumstance or the patient.

Another critical design factor is ease of use and maintenance. Many of the robotic devices are quite complex to use, with long setup procedures that may take up valuable allotted therapy time. To remedy this challenge, an effective design team should be represented by engineers, therapists, physicians, and patients. Similarly, physical aesthetics may play a large role in minimizing emotional barriers of use and should also be strongly considered in design decisions. Other development concerns include use of different materials such as soft and fluid material as opposed to traditional rigid structures, software security of the robots to prevent malicious hacking, and financial cost of the robotic system. Another important consideration is the ethics of robot use. To what extent do users have control over the robot? What about privacy of the data collected by the robots? These questions are universal as we adopt more robotics and computer systems in our lives.

Upper-Extremity Robotic Therapy

Hand and arm impairments are common sequelae of neurologic and musculoskeletal insults such as stroke, spinal cord injuries, Parkinson’s disease, and arthritis. Particularly after a stroke, there are a number of established physiotherapeutic approaches including Bobath, Brunnstrom, Proprioceptive Neuromuscular Facilitation, and task-specific training. Although none of these approaches show any significant advantages over the other, task-specific training has been the most common, sometimes used in combination with constraint-induced movement therapy. The goal of task specific training, also known as functional task practice, is to facilitate relearning of the appropriate motor skill to improve task performance. Given its acceptance and effectiveness, many of the upper-extremity therapy robots are based on repetitive task-specific motions.

Designing for upper-extremity therapy robots poses its own challenges. Shoulder joints are very complex, with 3 degrees of freedom and movement of the center of rotation while a shoulder is mobilized. The wrist also has three degrees of freedom and the hand with one to 2 degrees of freedom depending on the joint. Hand function, with the hand’s numerous joints, is usually one of the last places to recover, if at all, after a neurologic insult. Another difficulty is that upper extremities exhibit strong synergistic muscle contractions after a stroke, often leading to dystonic upper-extremity posturing, thus isolating individual muscle groups for exercise is challenging. Due to these concerns, conventional therapy focuses first on training the proximal limb such as shoulder and elbow.

Evidence for the use of upper-extremity therapy robots is mixed. Some findings imply that robotic-assisted therapy can facilitate motor recovery with meaningful changes in activities of daily living, but others suggest no difference from usual care. One important factor to consider is that the timing at which robot therapy is done also affects outcome. After a stroke, most of the recovery happens in the first 3 months. Studies demonstrate robotic therapy increases motor control in the chronic phase (beyond 3 months postinjury), whereas studies looking at acute/subacute phase (less than 3 months postinjury) cannot differentiate therapeutic intervention recovery from natural recovery.

For dose-matched trials that were performed for poststroke rehabilitation, when conventional therapy and robotic therapy with the same intensity and repetitions were compared, the outcome for robotic therapy appeared to be as effective as that of conventional therapy. Similarly, in robotic therapy after a distal radial fracture, there were no significant differences in outcome between robotic therapy and conventional therapy. As these robots were designed to simulate and reproduce therapists’ manual exercises, the comparable effectiveness is an argument for the implementation of robotic devices to offload therapist’s taxing tasks and allow the clinician to focus on functional rehabilitation.

Above all, there are no reported significant adverse events, and robot-assisted therapy appears to be safe. A limitation to the overall recommendation for robots in therapy is the heterogeneity of studies using different robots with different doses. Despite this variability, the evidence indicates that outcomes are independent of the use of a specific robot, but rather from the use of robots in general. This evidence has led the American Heart Association to announce that robotic therapy can be considered in delivering more intensive therapies.

Recent studies have shown that the use of robotics with botulinum toxin in spastic arm conditions has synergistic effects in improving upper-limb function. Additionally, some robotic therapy showed reduction in neuropathic pain in the shoulder and hands. These findings demonstrate that there may be other unexplored benefits to robotic therapy in the upper extremity.

Lower-Extremity Robotic Therapy

Gait dysfunction is another common sequela of neurologic and musculoskeletal injuries. Especially after a stroke, the ability to walk independently is one of the major determinants for home discharge, and walking speed is an indicator of functional community ambulation. Such as in upper-extremity rehabilitation, patients usually start their therapy course with passive range of motion, then move on to active exercises with assistance, active exercises without assistance, and active exercises with resistance. Conventional therapy for gait training involves varying combinations of strengthening, task-oriented exercises, functional electric stimulation, biofeedback, and treadmill exercises requiring many repetitive motions and many iterations of practice ; thus, gait training is a good candidate for robotic therapy.

For lower-extremity gait training, there are several different types of robots, including tethered and nontethered exoskeleton machines and end-effector type machines. In tethered exoskeleton systems such as Lokomat and LOPES, patients are usually suspended in a harness with robotic legs attached to their legs so that the exoskeleton supports and guides limb motion on a treadmill. There are untethered exoskeleton systems such as the ReWalk, Indego, and Ekso that allow for the most realistic walking experience. Details of exoskeleton devices will be mentioned later in this chapter. Examples of end effector–type robots include Gait Trainer, G-EO System, and GAR (see Fig. 6.3 ). These robots guide the patient’s feet through a gait cycle, imitating stance and swing phases with less friction and motor interference than exoskeleton devices. As such, they have greater variability in motion, which allow for balance training and stair climbing in addition to gait retraining.

For patients who are wheelchair bound, there are significant advantages to treadmill training with a system that supports body weight. Typically, two or three therapists are required to assist such patients not only to help support their weight, but to control trunk and limb movement. Replacement with a robotic system allows for early verticality, which improves cardiorespiratory rehabilitation and also stimulates wakefulness, an important goal especially in patients with severe cognitive neurologic impairment. Although gravity could be mostly eliminated through a body weight support mechanism, these robotic machines still require patients to have adequate endurance and the ability to stand. For those without those residual facilities, there are robots that allow training in a seated or lying position to work on building endurance and strength, mobility, and coordination.

As with upper-extremity robotic therapy, the evidence of robotic therapy over conventional therapy is mixed. For positive findings, patients who had stroke and who received robotic therapy in combination with conventional therapy seemed to have a higher likelihood of achieving independent walking. However, it is not clear whether exoskeleton devices or end effector–type devices are more effective or whether robotic therapy is effective for patients with subacute or chronic condition. Currently the strongest evidence is for robotic therapy with end effector–type devices, in conjunction with conventional therapy for patients with subacute condition after a stroke. Nevertheless, a number of studies did not show any significant benefit. Critics have noted that robotic devices like Lokomat are restricted in the coronal and axial planes, focusing only on movement in the sagittal plane and reducing training for postural stabilization. Additionally, although robots support patients and reduce cardiopulmonary load, this may inhibit the aerobic stimulus required for gait relearning. Finally, because robots assist in limb trajectory, there is little cognitive component, which is essential for neuroplasticity. Despite these mixed findings, the American Heart Association recommends considering robotic devices for poststroke therapy in patients who have limited ambulation or are nonambulatory.

For patients with spinal cord injury, a meta-analysis showed that robotic-assisted gait training improved endurance and mobility and decreased spasticity, but did not change the level of their pain. Authors proposed that rhythmic passive exercises from gait training could stimulate neuroplastic reorganization in the spinal cord, leading to decreased spasticity. Alternatively, critics also note that passive motion leads to degeneration of motor neurons with no patient’s cognitive input.

Virtual Reality in Rehabilitation

As mentioned previously, neurorecovery requires active engagement to promote neuroplasticity and motor relearning. However, the level of patients’ participation during therapy depends on their motivation and interest. One method of increasing popularity is to engage patients via virtual and augmented reality. Virtual reality technology creates an immersive environment for the patients beyond what is possible in the physical world. This technology has previously been used in other areas such as flight simulation, surgical training, treatment of phobias, and posttraumatic stress disorder.

There are many benefits from integrating virtual reality into therapy. First, patients describe the immersive environment as more interesting and thus encouraging higher intensity and repetitions during therapy. This can also be combined with games. Second, virtual reality could allow patients to practice daily activities beyond the hospital environment, such as in supermarkets and in the street, and individualize the training environment for the patient. And lastly, there are data that show virtual reality induces cortical reorganization and neuroplasticity.

With virtual reality, patients have reported greater motivation, less fatigue, improved visuospatial awareness, and overall better quality of life. This technique is especially helpful for patients with impaired cognition and reduced visuospatial awareness. Although particular task exercises may not be generalizable, improved visuospatial awareness will allow patients to adapt and participate in activities of daily living. Further, the visuospatial training for return to automobile driving could be done in a safe virtual environment. Virtual and augmented realities could also become another important tool for therapist assessment of patient performance, such as for automobile driving after stroke or for fall risks in those with Parkinson disease. In addition, pain management is another area that utilizes virtual reality by creating an illusion of the patient’s body, which may help with phantom limb pain.

Evidence for the use of virtual reality in therapy is still under investigation. There is some evidence to suggest a better ability to manage activities of daily living and better gait and balance after therapy utilizing virtual reality. Although virtual reality combined with robotic machines interfaces and tactile feedback could provide immersive and synchronized experience, many of the current technologies are too slow, causing a delay between responses in motor output and sensory feedback, which leads to increased task difficulty and motion sickness. The true effectiveness of virtual reality in rehabilitation is yet to be established and awaits better advancements in technology.

Social and Emotional Therapy Robots

Mental health is another area where robotic therapy can help. Patients with social anxiety or developmental disorders often have difficulty engaging with others and expressing and communicating emotions appropriately. Patients with autism spectrum disorders are naturally drawn to robots that have predictable behavior. There are robots that are socially interactive, participating in simple games with the patient for simple cognitive and physical exercises. One robot, Facial Automation for Conveying Emotions (FACE), is able to display and express lifelike human emotion to help patients with autism to identify and practice emotional expressions to improve communication skills. Other robots may look like a stuffed animal or a cute toy to increase interaction and companionship (see Fig. 6.4 ). Although these robots are specifically designed for social and emotional support, the concepts could also be applied to other therapy and assistive robots that are approachable and provide support to patients.

Assistive Robots

Despite intense rehabilitation, many patients still require long-term assistance, and with an aging population increasing demand for professional caregivers has led to a shortage. Fortunately, robots could fill this gap and moreover provide consistent and reliable care at any time of the day, whereas caregivers are limited by their working hours, as well as physical and mental fatigue. These assistive robots could decrease caregiver burden while improving patient autonomy and independence, thereby possibly delaying transition into nursing homes.

Assistive robots are designed to aid across various aspects of activities of daily living, including transfer and mobility, dressing, eating, hygiene, bathing, and toileting. Mobility aide robots such as nonanchored rolling frame robots can dynamically support the body and maintain a stable posture while allowing patients to walk around. Other robots have guidance system for ease of navigation, some with fully autonomous robotic wheelchairs that can drive patients around their house or even in the community . Transfers to and from beds to chairs are frequent and often lead to back pain in caregivers, but machines like the Robot for Interactive Body Assistance can provide assistance in carrying patients (see Fig. 6.5 ). Other tasks requiring more fine motor control such as feeding, bathing, and toileting are better facilitated by articulated manipulator arm robots with 3 or more degrees of freedom. These manipulators can be mounted on a surface, and through artificial intelligence, patients can command the arm to do tasks such as grasping, pushing, and brushing. A particular commercialized robot, Obi, is specifically designed for feeding (see Fig. 6.6 ).