Abstract
Background The wrist extensors and flexors are profoundly affected in most children with hemiparetic cerebral palsy (CP) and are the major target of physical therapists' and occupational therapists' efforts to restore useful hand functions. A limitation of any therapeutic or exercise program can be the level of the child's engagement or adherence. The proposed approach capitalizes on the primary learning avenue for children: toy play.
Objective This study aimed to develop and evaluate the measurement accuracy of innovative, motion-specific play controllers that are engaging rehabilitative devices for enhancing therapy and promoting neural plasticity and functional recovery in children with CP.
Design Design objectives of the play controller included a cost-effective, home-based supplement to physical therapy, the ability to calibrate the controller so that play can be accomplished with any active range of motion, and the capability of logging play activity and wrist motion over week-long periods.
Methods Accuracy of the play controller in measuring wrist flexion-extension was evaluated in 6 children who were developing in a typical manner, using optical motion capture of the wrist and forearm as the gold standard.
Results The error of the play controller was estimated at approximately 5 degrees in both maximum wrist flexion and extension.
Limitations Measurements were taken during a laboratory session, with children without CP, and no toy or computer game was interfaced with the play controller. Therefore, the potential engagement of the proposed approach for therapy remains to be evaluated.
Conclusions This study presented the concept, development, and wrist tracking accuracy of an inexpensive approach to extremity therapy that may have a health benefit for children with hemiparesis, and potentially for patients of any age with a wide range of extremity neuromotor impairments.
Cerebral palsy (CP) is a neuromotor impairment comprising a group of nonprogressive clinical syndromes of children that are characterized by motor and postural dysfunction. In the United States, CP affects approximately 3.1 out of every 1,000 children, and each year more than 10,000 infants and children are diagnosed with this condition.1,2 The wrist extensors and flexors and forearm supinator and pronator muscles are profoundly affected in most children with hemiparetic CP and are the major target of physical therapists' and occupational therapists' efforts to restore useful hand functions.3 The need for physical therapy is especially important during motor development, as children strive to master more difficult motor tasks and deal with secondary effects such as bone deformities because of contractures.
Physical therapy, occupational therapy, braces, orthotics, electrical stimulation, medications, and surgery are the major forms of treatment for hemiparetic CP. These treatments are intended to improve strength and range of motion (ROM), prevent contractures, and increase function. Although these approaches primarily target muscle symptoms, approaches to reshape the nervous system are now gaining attention because it is established that the nervous system, especially in the developing brain, is capable of reorganization and change through a variety of “neural plasticity” mechanisms.4 Although recently published studies suggest that therapy improves functional outcome,5,6 the most effective amount, intensity, timing, frequency, and duration to affect the neural system have not been established. The overriding consensus is that more therapy equals a better outcome.5,6
Constraint-induced movement therapy (CIMT) is a form of rehabilitative therapy that limits use of the unaffected limb through casting or other orthotic devices to promote use and practice of “natural” activities of daily living with the affected extremity. The concept behind CIMT is to engage use-induced plasticity to allow relearning of neuromuscular control in weak or spastic limbs.7,8 An important aspect of CIMT is the promotion of increasingly advanced levels of performance with the impaired upper extremity.9 This shaping procedure involves providing feedback for small improvements in task performance, lowering the threshold of a movement sequence if the individual is incapable of completing the movement on his or her own at first, and systematically increasing the difficulty of the task performed.10 Robotic therapy is another form of rehabilitative therapy that generally uses advanced computer-controlled exoskeletons to facilitate desired upper extremity movement. This therapy has largely been developed for use in stroke therapy for adults11,12 and has been found to improve functional outcome measures.13 Early-stage results with robotic therapy suggest a benefit in children with CP,13,14 even though training was limited to 18 one-hour sessions over 3 weeks.
Our long-term goal is to develop and evaluate innovative motion-specific play controllers that are engaging rehabilitative devices for enhancing therapy and promoting neural plasticity and functional recovery in children with CP. Our approach combines the ongoing practice of CIMT and the quantitative assessment of robotic systems while addressing the limitations of constraining the more functional limb and the high cost and institutional limitations of robotic therapy.
The design objectives of the play controller included a cost-effective, home-based supplement to physical therapy, the ability to calibrate the controller so that play can be accomplished with any active ROM, and the capability of logging motion over long play periods in order for us to evaluate its efficacy and the potential for a dose response. The aim of this study was to evaluate the accuracy of the play controller in measuring wrist flexion and extension in a cohort of children who were developing in a typical manner. Motion tracking accuracy was established using optical motion capture of the wrist and forearm as the gold standard.
Method
Play Controller Design
The play controller was designed with a conformable, ergonomic handle to accommodate varying levels of contractures among children with CP. The controller is composed of 4 main components: a foam handle, plastic wrist hinge, soft fabric forearm cuff, and electronics closure (Fig. 1). The handle is composed of closed-cell foam tubing (2.54-cm [1-in] diameter, lengths vary with customized sizing) wrapped over a ductile aluminum alloy wire (0.64-cm [0.25-in] diameter, lengths vary with customized sizing). The ductile wire connects the foam handle to the wrist hinge and, because it is easily malleable, allows the length and the orientation of the handle to be easily customized to the size and posture of each child's wrist. The single-axis wrist hinge was 3-dimensionally printed, was made out of acrylonitrile butadiene styrene plastic (25-mm diameter) material, and houses a potentiometer to measure wrist flexion and extension. The wrist hinge is rigidly attached to a thin plate of spring steel that is embedded within the fabric forearm cuff. The fabric that makes up the forearm cuff and straps is breathable, open-cell foam bonded to a smooth nylon tricot on one side and a high-nap fabric on the other side to accommodate hook-and-loop fasteners (AirFlex, Eastex Products Inc, Holbrook, Massachusetts). The forearm cuff is secured to the child's forearm by 2 straps attached with hook-and-loop fasteners. A fabric strap assists in securing the child's hand to the grip. The electronics of the controller are encased within a 3-dimensionally printed plastic shell (82 mm × 75 mm × 34 mm) and attached to the forearm cuff with a hook-and-loop fastener.
Play controller components consist of handle with malleable rod; wrist joint housing the potentiometer; forearm cuff of soft, breathable, self-adhesive matting material over a malleable fitting cuff; and the electronic housing.
The controller was designed to interface with commercially available toys, provide selectable thresholds for play ROM, and log data of motions and durations of play. The controller electronics included a 32-bit, 48-MHz ARM Cortex-M3 microcontroller (SAM3S4B, Atmel, San Jose, California); a rotary potentiometer with linear taper (10-kΩ linear taper, 300°±5°), a triaxial accelerometer (±3 g, ADXL335, Analog Devices, Norwood, Massachusetts); a micro SD card, and a Bluetooth 4.0 module (BR-LE4.0-S2, BlueRadios Inc, Englewood, Colorado) (Fig. 2). The controller sampled the potentiometer and logged data to a 4-GB micro SD card whenever the potentiometer moved more than 4 degrees. The data log included the time of measurement, measured values from the potentiometer, and measurements from the triaxial accelerometer, which were not evaluated in this study but could be used to calculate forearm orientation (pronation and supination). The wrist hinge of the play controller was designed to protect the potentiometer from overrotation and, therefore, was intended to limit wrist extension and flexion to ±90 degrees, for a total ROM of nearly 180 degrees. The average maximum extension and flexion angles for the mechanical limits of the fabricated play controllers (n=10) were 87.7 degrees (SD=2.5) and 86.5 degrees (SD=2.8), respectively. Figure 3 shows the play controller fitted to a child's hand and a toy car that the child can operate with the controller. For the controller to operate the car, the electronics of the car are replaced with a similar module to allow control via Bluetooth for an extended wireless range beyond the typical capability of the toy car.
Three-dimensionally printed electronic housing (H1 and H2), control buttons (F), battery (A), and electronic boards (B, C, D and E) for the play controller.
The remote control car shown here is purchased with a commercial controller that requires thumb motion to drive forward or backward. When direction is reversed quickly, the car can flip and continues driving. The steering is not controlled but rotates freely. The specially designed play controller replaces the commercial thumb controller, is readily custom fitted to the child, and drives the car forward and backward with wrist flexion and extension motions. The positions of wrist flexion and extension that trigger the car's change in direction is programmable for each individual child and is set by holding a button on the electronics enclosure.
Play Controller Accuracy
After institutional review board approval, 6 children (3 male, 3 female; age range=5–11 years) who were developing in a typical manner were recruited to participate in this pilot study. The children were fit with the play controller and their arms were outfitted with retroreflective markers. The children were seated in view of the optical motion capture system (4 Oqus 5-series infrared sensing cameras, Qualysis, Gothenburg, Sweden), positioned with their arm in a neutrally pronated position, and instructed to bend their wrists back and forth between flexion and extension. It should be noted that children were not instructed to try to reach their maximum flexion or extension angles; they were simply instructed to cycle their wrist through flexion and extension for a short period of time.
Wrist flexion and extension angles were tracked at 250 Hz using a cluster of markers on the hand and one marker on the distal forearm. Hand and forearm segments were defined using anatomical markers placed on the radial and ulnar styloids, medial and lateral epicondyles of the humerus, and first and third metacarpals. Voltage from the potentiometer mounted in the play controller was sampled by the Qualysis analog system at 250 Hz and converted to degrees of rotation using customized MATLAB code (The MathWorks Corp, Natick, Massachusetts). Wrist angular positions from the optical markers were determined using Visual3D (C-Motion Inc, Germantown, Maryland). Wrist motion was broken down into cycles, where one cycle was defined as the transition from peak extension to peak flexion. Full ROM, maximum flexion, and maximum extension angles for both the controller and the motion capture system were calculated for each cycle. Absolute errors of the play controller joint angle measures were calculated as the average difference between the recorded angles and the motion capture system for each cycle and expressed in degrees and as percentages. Differences in the time of maximum joint angle between the 2 systems were reported.
Data Analysis
The Bland-Altman method15 was used to evaluate differences in joint angles collected by the optical motion capture system and the play controller. Additionally, Spearman correlation coefficients (as the measures were not normally distributed, Shapiro-Wilk test; P<.05) and P values for measurements of maximum flexion and extension between the 2 systems were calculated.
Role of the Funding Source
The research reported in this publication was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health under award number R21HD071582. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors are indebted to Richard Maddocks and his associates at Hasbro Inc, Pawtucket, Rhode Island, for their support.
Results
We were able to successfully log data and analyze 93 wrist motion cycles. The average error between motion capture-derived ROM and play controller-recorded angles for all cycles was 10.1% (8.4°). For extension, the average error associated with peak values for all cycles was 4.9% (4.1°). Similarly, the average error associated with peak flexion values for all cycles was 5.2% (4.3°). Controller maximum flexion and extension angles occurred, on average, 0.07 seconds after the optical motion capture calculated the maximum joint angles.
Flexion values were found to have a bias (1 SD), or average discrepancy between methods, of −0.5 degrees (SD=5.4) and 95% limits of agreement from −11 degrees to 10 degrees (Fig. 4A). Extension values had a bias (1 SD) of −0.4 degrees (SD=5.1) and 95% limits of agreement from −10 degrees to 10 degrees (Fig. 4B). In both directions, the bias of the play controller was an underestimate of the optical motion capture system values. The correlation coefficient for maximum flexion values between the 2 systems was .86 (P<.001). For maximum extension values, the correlation coefficient was .88 (P<.001).
Bland-Altman plots for the difference in maximum flexion (A) and maximum extension (B) measurements between play controller and optical motion capture system. The x-axis represents the mean maximum flexion (A) and extension (B) measurements from the controller and the optical motion capture system. The y-axis represents the difference between maximum flexion (A) and extension (B) measurements from the controller and motion capture system. Negative y-axis values indicate that the controller underestimated the optical motion capture system values. Bias lines and 95% confidence limits are represented as dashed lines.
Discussion
In this study, we sought to describe the development of a novel play controller and to evaluate how well it was able to track and record wrist flexion and extension motion. We found that the error, when compared with the optical motion capture system, was approximately 5 degrees in both extension and flexion. We anticipate that this accuracy would be sufficient to detect a clinically important difference in wrist motion with future interventional studies, but this remains to be demonstrated.
The average discrepancy between methods, or bias, for maximum flexion and extension was −0.5 degrees and −0.4 degrees, respectively. The width of limits of agreement for maximum flexion and extension was approximately 20 degrees, or 10% of the full ROM of the play controller (180°). The observations were centered on 0 degrees, and there was no statistical systematic variation of the differences with the mean. Therefore, the error between the systems is best summarized by the standard deviation of the differences (ie, 5.4° for flexion and 5.1° for extension). As shown in Figure 4A, there appeared to be a bias at large flexion angles, which was not statistically significant and may have been due to the controller sliding on the forearm at these large flexion angles. Nonetheless, the sufficiently small errors, from a practical or clinical standpoint, can be interpreted as minimal differences existing between the play controller measurements and the motion capture system. The correlation coefficients calculated for maximum flexion and extensions (.86 and .88, respectively) were in agreement with the Bland-Altman plots; measurements of joint angles between the play controller and the motion capture system were strongly correlated.
Although the Nintendo Wii (Nintendo, Redmond, Washington) and other similar gaming technologies are clearly potential tools for rehabilitation,16,17 these technologies are based on measurements of global controller motion, not specific joint motion. For example, if tilting the Wii controller is the required motion to play the game, this tilt could be accomplished with any nonspecific hand, forearm, or even torso movement. In other words, these gaming technologies track only terminal movement of the controller regardless of how the limb is positioned in space. The strength of our play controller is that it measures wrist flexion and extension regardless of posture of the upper limb. The approach is easily applicable to joints other than the wrist and forearm.
This initial evaluation study had several limitations. The measurements were taken during a laboratory session in which no toy or computer game was interfaced with the play controller. Therefore, the potential engagement of the proposed approach for therapy remains to be evaluated. Although we do not believe that the measurement accuracy of the play controller would be different in a cohort of children with CP, this study was conducted only on children who were developing in a typical manner.
This report introduces and evaluates the measurement accuracy of a therapeutic device for children that is designed to combine the benefits of CIMT and robotic therapies with cost-effective home delivery. A limitation of any therapeutic or exercise program can be the level of the child's engagement or adherence, particularly if an intervention is perceived as “boring.”18 The proposed approach capitalizes on the primary learning avenue for children: toy play.19 Using play as a motivator, we propose that we can maximize therapy adherence, a significant factor in improved outcomes.9 Moreover, our approach has a potential secondary benefit to quality of life. Children with physical disabilities who have not been able to engage in “typical” play activities may experience secondary social and emotional disabilities.19 Play contributes to the physical, cognitive, social, language, and emotional development of children.20,21 Greater involvement in play and leisure activities is associated with improved coping and psychosocial functioning for children with congenital physical disabilities.22
In our approach to develop a therapeutic device, we have attempted to combine (1) targeted joint motion; (2) massed practice therapy; (3) logging of play and therapy activity (a “dosage meter”); (4) pediatric focus; (5) play as motivation for therapy; (6) use of existing, commercially proven toys and games; and (7) cost-effectiveness. Although whole-body exercise has clear benefits, repetitive targeted therapy of specific muscle groups, especially in the hemiparetic hand and wrist, is essential in all stages of a rehabilitation program.23 Upper extremity function often becomes limited with progressive restriction of movement due to contractures about the wrist and elbow. A joint-specific play controller may be advantageous because it is focused on repetitive tasks of specific wrist and forearm motion to promote more direct effects on these muscles, in addition to promoting changes in the central nervous system.
In conclusion, this study presented the development and measurement validation of an inexpensive approach to extremity therapy that may have a significant health benefit for children with hemiparesis and potentially for patients of any age with a wide range of extremity neuromotor impairments.
Footnotes
Dr Crisco, Mr Schwartz, Dr Wilcox, and Dr Kerman provided concept/idea/project design and writing. All authors provided data collection. Dr Crisco, Mr Schwartz, and Dr Wilcox provided data analysis. Dr Crisco, Mr Schwartz, and Dr Kerman provided project management. Dr Crisco and Dr Kerman provided fund procurement, facilities/equipment, and institutional liaisons. Dr Kerman provided participants and administrative support. Mr Schwartz and Dr Kerman provided consultation (including review of manuscript before submission).
This study was approved by the Institutional Review Board of Rhode Island Hospital.
The research reported in this publication was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health under award number R21HD071582. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors are indebted to Richard Maddocks and his associates at Hasbro Inc, Pawtucket, Rhode Island, for their support.
Dr Crisco is an inventor on a patent describing a method for facilitating fitting of the play controller to a child's arm using a malleable inner structure. The patent is owned by his employer, Rhode Island Hospital, Providence, Rhode Island.
- Received August 4, 2014.
- Accepted January 2, 2015.
- © 2015 American Physical Therapy Association
References
Issue highlights
