Home-Based Functional Electrical Stimulation-Assisted Hand Therapy Video Games for Children with Hemiplegia: Development and Proof-of-Concept

Abstract

We describe the development and three case reports of a home-based intervention for children with hand hemiplegia that integrates custom video games with contralaterally controlled functional electrical stimulation (CCFES). With CCFES, stimulated opening of the more-affected hand is modulated by volitional opening of the less-affected hand. Video games that solicit goal-oriented, skill-requiring movement have shown promise for treating hemiplegia, but they have not previously been combined with electrical stimulation in children. Three children ages 8, 9, and 11 with moderate-to-severe hand hemiplegia were assigned six weeks of therapy in lab and at home. The goal was to determine if children could tolerate 9 lab treatment sessions and administer up to 7.5 hrs/wk of CCFES video game therapy at home. The feasibility of this intervention for home use was assessed by device logs, end-of-treatment interviews, and motor function/impairment assessments. With caregiver help, the children were all able to attend 9 lab sessions and built up to 7.5 hrs/wk of therapy by week 3. They averaged 5-7 hrs/wk of home intervention overall. Motor outcomes improved for all three participants at treatment end, but mostly regressed at 4-weeks follow-up. Individual improvements at treatment end exceeded minimum detectable or clinically important thresholds for Assisting Hands Assessment, Fugl-Meyer Assessment, and Melbourne Motor Assessment 2. We found preliminary indications that CCFES-integrated video game therapy can provide a high dose of hand motor control therapy at home and in the lab. Improvements in motor outcomes were also observed, but more development and study is needed.

I. Introduction

HEMIPLEGIA affects approximately one third of children with cerebral palsy [1]. Treatment approaches based on intensive bimanual therapy and constraint-induced movement therapy (CIMT) can reduce upper-limb impairment, but children with insufficient volitional movement for practicing activities of daily living may not be able to participate or benefit from these treatment modalities. Therapies promoting hand use for children with varying ability are needed – especially at home, where non-use of the more-affected, paretic hand is high and adherence to home therapy programs is low [2].

Our intervention combines custom video games with contralaterally controlled functional electrical stimulation (CCFES) to facilitate practice using the more-affected hand (Fig. 1). These are complimentary techniques that have shown promise when used independently to treat hemiplegia in children, but have never been combined for pediatric intervention. CCFES uses volitional opening of the less-affected hand to control the intensity of stimulation to the paretic extensor muscles of all four fingers and the thumb. This novel electrical stimulation modality for motor relearning was originally developed for and been shown to improve hand opening, dexterity, and upper limb impairment for adults with hemiplegia after stroke [3]–[7]. CCFES may also have the potential to benefit children with cerebral palsy, but it has not yet been applied to this population. The goal of the current study was to explore the suitability of this intervention for children with hemiplegia and identify necessary steps to make the intervention more viable for children. The key research question was to investigate whether children with hemiplegia are able to tolerate an intensive intervention using CCFES-assisted hand therapy video games. The objective was to conduct a proof-of-concept study of three children receiving a six-week intervention consisting of nine lab visits and up to 7.5 hrs/week of home use.

Fig. 1.

Child with hemiplegia using CCFES to assist the more affected right hand to open during hand therapy video game training.

This article describes: a) the rationale for the new treatment, b) the design of CCFES-compatible video games, and c) case reports from an exploratory home-based therapy regimen that was trialed in three children with hemiplegia.

II. Background

The clinical effectiveness of FES for cerebral palsy requires more evidence, [8], [9], but case studies found that FES benefits hand impairment mechanisms. Atwater, et al. found that wrist extensor FES triggered by electromyography increased wrist range of motion after 8 weeks (3 hours/week) [10]. Wright and Granat applied cyclic FES to the wrist extensor for 6 weeks (30 minutes/day) and found increased range of motion, but not extension torque. However, Kamper, et al. reported increased wrist extension torque, possibly due to concurrent decrease in wrist flexor spasticity, after applying cyclic FES to both the wrist flexor and extensor muscles for 12 weeks (15 minutes/day) [11].

Although beneficial, conventional FES is challenging to integrate with a broad range of manual tasks or motion-controlled video games (which often require partial hand opening) because it provides only full, not graded paretic hand opening assistance. The timing and intensity of cyclic FES is pre-set, so a child with hemiplegia does not control when the stimulation turns on/off or the amount of hand opening it produces. Electromyography-triggered FES opens the hand when muscle activation exceeds a threshold, but the child cannot control when stimulation turns off, nor the intensity and degree of hand opening. Switch-triggered FES also does not allow the child to grade the amount of hand opening.

Contralaterally-controlled FES was developed to provide people with hemiplegia graded control of paretic hand opening assistance to promote paretic limb use during task practice [3]. With CCFES, a glove instrumented with kinematic bend now.sensors (Fig. 1) is worn on the less-affected hand to translate hand opening and closing into proportional intensities of surface stimulation of the extensor muscles of all four fingers and the thumb (extensor digitorum, extensor indicis, extensor pollicis brevis/longus). Thus, volitional opening of the less-affected hand produces a proportional amount of stimulated opening of the more-affected hand. This assists individuals with hemiplegia to use their more-affected hand to perform tasks that require graded control of hand opening. Participants are instructed to open both hands at the same time while practicing functional tasks in order to produce bilateral cortical motor activity. This approach allows people with hemiplegia to practice using their paretic hand without requiring any residual movement or electromyogram signals from the more-affected, paretic upper extremity. Several studies by our group showed that the key features of CCFES (pairing motor intention to motor response, bilateral symmetric movement, and stimulation-assisted practice of goal-oriented tasks) have promise for treating hemiplegia for adults with chronic (> 12 months) [6] and subacute (< 6 months) [5] stroke. Most notably, adults who received CCFES therapy experienced greater reduction in hand impairment and activity limitation than participants who received cyclic FES therapy (which repetitively stimulated the paretic hand without a link to motor intention or voluntary muscle contraction from either limb) [5], [6]. Others have also reported evidence from randomized control trials that CCFES has benefits over cyclic FES for active range of motion, reduced impairment, and improved upper extremity function for adults < 15 days post-stroke [12], < 3 months post-stroke [13], and > 6 months post-stroke [14]. Stimulating the hand extensors is expected to support improved hand function because it strengthens muscles and makes them more fatigue resistant [15]. We found evidence in adults with stroke that CCFES, unlike cyclic FES, can promote cortical changes that facilitate recovery, ipsilesional or contralesional excitability and inter-hemispheric inhibition [16].

Video game features may enhance CCFES treatment. Our prior studies trialed 10 sessions/week of repetitive CCFES-mediated repetitive hand opening exercises lasting 45 minutes per session at home in addition to two 90 minute sessions per week of functional task practice with an occupational therapist in the lab. Therefore, 70% of the dose consisted of hand opening exercises at home, where adults with stroke were prompted to repeatedly open both hands for several seconds in response to auditory cues. Repetitive hand opening facilitates mass practice, which is important for motor recovery, but it does not require significant attention, motor planning or goal-oriented motor skill practice, which are also important [8], [17], [18].

Unlike adults with stroke, children with hemiplegia are not likely to adhere to such a monotonous task when not monitored by a therapist in the home environment. In fact, the first child with hemiplegia who was recruited for our current study refused to adhere to a regimen of repetitive hand opening in response to beeps for 45 minutes/day, 10 times/week.

We hypothesized that replacing the repetitive hand opening exercise with goal-oriented CCFES-assisted video games would encourage therapy adherence. Custom video games have been developed by us to maintain attention on practicing goal-oriented hand tasks that require motor planning and skilled motor control. Adults with stroke were able to self-administer the combined CCFES with video game treatment at home with very good adherence [7], but it is unknown if children with hemiplegia will adhere to this intervention at home. It is also important to investigate which games appeal to children and if the combination of electrode placement, stimulator operation, and game use is too complex for children and their caregivers.

Custom virtual environments are a promising modality for neurorehabilitation in adults with stroke [19] and children with cerebral palsy [20]–[24]. Although positive outcomes can be achieved with conventional motion-controlled video games that were not designed for rehabilitation [8], [25], [26], there is evidence that those developed specifically for rehabilitation led to greater motor outcome improvements [21]. Conventional games that are too difficult may hinder rehabilitation by eliciting and reinforcing abnormal movement patterns [27], [28] or be unsuitable for children with more severe impairment.

Paretic limb motion is typically used as the input for video games or virtual tasks that have been designed around principles associated with effective motor learning [29], [30]. Such principles include repetition of goal-directed movement, performance feedback about the trained skill or task [31], adjustable difficulty level appropriate to ability [27], [32], reduced effect of compensatory movements on gameplay success [33], and increased motivation using performance feedback/achievement features found in conventional video games [34]. Although the clinical effectiveness of virtual environments requires more evidence for children with cerebral palsy and large randomized trials are few – studies have consistently shown positive pre- and post-treatment effects [8], [21]. The current consensus from reviews [8], [20], [22] and meta analyses [21] is that virtual environments are low risk and have sufficient evidence to be included as an adjunct to conventional therapy for children with cerebral palsy.

Combining CCFES with hand therapy video games is novel and can enable them to practice tasks that are otherwise not possible. Rehabilitation robots have been paired with video games [35]–[38], but robotics are not yet viable for home use or assisting activities of daily living. In contrast, CCFES is portable and can assist during activities of daily living or video games that require fine control of hand-opening.

III. Methods

A. Game Development Process

Four video games were developed for CCFES. During development, each game was evaluated by neurorehabilitation experts, including 3 physiatrists, 2 therapists, a biomedical engineer at MetroHealth, and participants in ongoing CCFES studies with post-stroke hemiplegia. Each of them played the games and gave feedback on whether the games exhibited all the following features important for motor relearning.

• Engaging and Goal-Oriented: Game-like features, such as conflict, competition, level progression, ability to win, are used to motivate repetitive paretic hand use [29], [30], [34].

• Skill-Focused: In-game success requires participant to focus on skilled, precise control of hand opening [33].

• Intuitive Control: Game is easy to learn and provides a close match between physical and virtual movement, which also discourages use of compensatory movements to play.

• Suitable Difficulty: Gameplay difficulty can be adjusted or shaped to provide an appropriate range of challenge for a wide spectrum of hand ability [27], [32].

• Provides Relevant Feedback: Performance feedback provided to participants motivates desire to improve game performance and motor skill through error correction, replay, and repetitive optimization [31].

• Parsimonious Presentation: Images and sound are suitable for people with mild cognitive or visual impairment.

• Gameplay control requires only hand opening/closing.

Each game was repeatedly revised based on their suggestions. After development, the suitability of the games for home and lab use were studied in 3 adults with post-stroke hemiplegia (reported in [7]) and 3 children with hemiplegic cerebral palsy (the current study). Consistent with game design best practices [39], knowledge gained from these studies will inform future refinement of our intervention for for people with hemiplegia, their families, and occupational therapists.

B. Video Game System

All the video games were designed using the Unity game development engine (Unity Technologies, San Francisco, CA). As shown in Fig. 1, the games ran on a 24 inch touchscreen desktop computer (Dell Inspiron 2330, Dell Inc., Round Rock, TX) running Windows 8 (Microsoft Corp., Redmond, WA). For all the games, the child wore a bend sensor (Images SI, Inc., Staten Island, NY) on the more-affected hand. The sensor was attached to a fingerless mitt (Handana, Austin, TX) that had an opening for the thumb and a larger opening for the four fingers (Fig. 1), making it easy to wear even when the fingers cannot be fully extended due to contractures or muscle tone. A single bend sensor was secured to the dorsal aspect of the index finger using 3D printed rings that were sized to each participant. These rings kept the sensor straight regardless of hand position.

An analog-to-digital converter (Phidgets 8/8/8 Interface Kit, Phidgets Inc.) sampled bend sensor data at 1 kHz. The analog-to-digital converter was housed in an enclosure (Fig. 2 right) which interfaced to the PC via USB connection. The rationale for using bend sensors was to maximize reliability and minimize electronic footprint on the paretic hand. Wireless inertial and optical sensors were considered, but inertial sensors can drift and wireless sensors require risk battery or connectivity failures and have a larger electronic footprint. Bend sensors used in this study have been reported to have high linearity, reliability, and repeatability for joint sensing [40].

Fig. 2.

(Left) CCFES device consisting of surface stimulation electrodes placed on the extensor muscles of the four fingers and thumb of the paretic arm, stimulator, and sensor glove worn on the non-paretic hand. (Right) Touchscreen computer running the virtual tasks and games controlled by a bend sensor worn on the paretic hand with a fingerless mitt. The Game Menu with all four game launching icons are shown on the touch screen.

Games used in the lab were configured to “therapist mode” and the ones at the participant’s home were configured to “participant mode.” Therapist mode allowed study staff to restrict the list of games displayed on the main menu to those that were assigned and to adjust game difficulty. Therapist mode saved the difficulty settings configured by staff into an XML computer data file stored on a portable USB memory stick that has to be connected for the games to run. When children start a game at home, it loads the difficulty settings automatically from the XML computer file on the USB memory stick. In participant mode, the games did not allow children to change difficulty settings. Games also stored usage data on the USB memory stick and families were instructed to bring the USB stick to lab visits so we could download the logs.

C. Video Game Descriptions

The Game Menu allowed participants to touch and launch games from a list of those assigned by the treating therapist (Fig. 2 right). The icons of games that were not assigned were hidden using the therapist mode of the Game Menu. The Game Menu also enforced a 3-minute rest period between games by displaying a countdown timer and re-displaying the game selection icons after the rest. Table I describes game details.

TABLE I.

Hand Therapy Video Game Descriptions

	Paddle Ball (Fig. 3 upper left)	Sound Tracker (Fig. 3 lower left)	Skee Ball (Fig. 3 upper right)	Marble Maze (Fig. 3 lower right)
Goal	Deflect the ball past computer opponent’s paddle and score 7 points.	Guide a cursor to trace a track generated from frequency spectrum changes in a song.	Launch a ball into the target bullseye, which is repositioned vertically after each hit.	Rotate a maze hanging on the wall to guide marbles to fall out and into a bucket.
Skill Focus	Graded, timed movement across range of motion.	Precise, continuous control of graded hand opening [35]–[37].	Precise control of graded hand opening speed.	Sustain varying degrees of hand opening for marbles to roll.
Control	Hand opening angle proportionally controlled vertical paddle position. Ball speed and deflection angle could be boosted by contacting paddle edges or by moving paddle faster to “chop” at the ball. Participants controlled the paddle on the same side as their more affected hand.	Hand opening angle proportionally controlled the vertical position of the cursor. Track amplitude was calibrated to vary from 20-80% of the paretic hand’s CCFES-assisted range of motion. Track and paddle motion was matched to the same side as the more affected hand.	Hand opening moved a block forward to strike and launch the ball to a distance proportional to the paretic hand’s maximum opening speed. A calibration taking the average of 3 fastest hand openings made the ball reach the top of the backboard with maximum CCFES-assisted hand opening speed.	Hand opening angle proportionally rotates the maze. Right hand opening rotated the maze counterclockwise and vice versa. A calibration mapped max hand opening to 380° of maze rotation and fully closed hand to its initial orientation, providing fine rotation control of about 2 degrees resolution.
Difficulty	Paddle width was set to ensure the paretic hand could move paddle’s edges to the top/bottom of the screen. Ball speed was set so children could deflect 75% of balls. Computer-controlled opponent skill level (junior varsity, varsity, and pro) was set so the child won most of the games.	Children were told to follow the track with the red center of the cursor, but the yellow cursor edges widened so participants could reach the top and bottom of the screen and the track was widened so they could complete each track without rest.	Target ring size was changed manually by study staff so that participants could obtain 70-80% accuracy (percentage of hits relative to the number of target strikes plus misses). The bullseye in the center of the target did not change in size.	Marble speed was adjusted so children could complete each maze. Mazes of increasing complexity were introduced after completing a maze with 10 concurrent marbles. Concurrent marbles were added when children beat a completion time goal set by based on the number of marbles and maze complexity.
Feedback	Immediate feedback was provided via a score board. After each 15-minute session, accuracy (% of hits divided by number of hits and misses) and the number of games won and lost were displayed to emphasize accuracy and skill as the focus of the game.	Immediate feedback was provided by increasing music volume when the cursor was on the track and changing the track green when it contacted the cursor center and yellow when it contacted the edges of the cursor. At the end of a session, tracking accuracy (% of track length followed relative to the total track length) was displayed.	A 15-minute session was split up into 3 5-min games and the score from all 3 games were displayed. Bullseye awarded 200 points vs. 100 points for the rest of the target. Immediate feedback included the number of ball launch attempts, points scored for all three games, high score, and speed of the current attempt was shown on a meter as % of max.	Immediate feedback included completion time, fastest prior completion time (unique for each maze and number of marbles), number of concurrent marbles, and the maze level number. After each 15-minute session, accuracy (% of marbles in the bucket vs the total number of marbles that exited the maze) was displayed.
Presentation	The ball was a distinct color from the paddles and walls for easy differentiation. Sounds effects were created for paddle strike, wall ricochet, and scored goals.	A white track moves toward the cursor, which changes colors as described in the feedback details. Songs that were used to generate the tracks were all public-domain instrumental melodies.	The virtual skee ball track has only one target. The speed meter indicated where the ball will fall on the backboard and a chime for bullseyes was higher pitched than for hitting the rest of the target.	The marbles, maze walls, and moving obstacles were distinct colors. Rolling sound effects increased in pitch with marble speed and a chime sounded when marbles entered the bucket.
Data logged	Paddle size, ball speed, goals scored, misses, number of games won, points scored.	Accuracy, tracking cursor size, and track width.	Accuracy, high score for each session, target ring radius, and maximum hand opening rate.	Marble speed, bucket size, and completion times.

All games included an automated calibration routine that prompted them to open and close the more-affected hand with CCFES assistance. This recorded the maximum and minimum bend sensor value ranges and hand opening speed from the more-affected hand. The calibration could be repeated as needed during gameplay to accommodate for changes in muscle tone or fatigue. children were trained to perform the calibration themselves before playing each of the four games. They were also instructed to re-calibrate whenever they noticed an inability to move the game object under their control to its minimum or maximum position (for Paddle Ball, Sound Tracker, Skee Ball) or rotation (for Marble Maze). Each game also had a pause button that we instructed participants to use in cases where the hand did not respond at all to CCFES. In this study, 1-3 minutes of rest typically restored full hand opening.

In addition to each game’s unique data log (Table I), common data elements logged for all games included: play duration, calibration minimum and maximum range of bend sensor values, and the number of hand motion repetitions. one hand motion repetition was defined as two sign changes in the first difference of the paretic hand’s bend sensor value.

D. Stimulation Device

Up to three transcutaneous electrodes (PALS, Axelgaard Manufacturing Corp.) were placed on the skin surface over the extensor muscles (extensor digitorum, extensor indicis, extensor pollicis brevis/longus) of all four fingers and the thumb to fully open the more-affected hand of each participant.

Stimulation was applied by a custom battery-operated transcutaneous stimulation unit developed at the Cleveland FES Center (Fig. 1) [41]. Biphasic current-controlled stimulation was delivered 35 Hz with 20 mA amplitude from up to 3 independent channels. Stimulation intensity was modulated using pulse widths between 0-250 μs.

The stimulator was programmed by study staff to achieve pain-free paretic hand opening using a Labview software interface (National Instruments, Inc.). First, a maximum stimulation pulse duration was determined that produced a functional degree of extension of all four fingers and the thumb without pain. Then the stimulator was programmed to transition from zero to maximum pulse duration for each electrode in proportion to the command glove’s amount of opening.

E. Participants

Three children (Table II) were recruited from outpatient therapy populations of Cleveland Clinic Children’s hospital for Rehabilitation and their caregivers provided written assent and consent, respectively, to participate in an exploratory case series study approved by the Cleveland Clinic IRB. Inclusion criteria were ages 6-12 with cortical lesions and hemiparesis by age two, 30-minute recall of two-out-of-three words, paretic extension of all four fingers strength scale ≤ four-out-of-five [42], ability to follow three-stage commands, sufficient shoulder/elbow range-of-motion for task practice, intact upper-limb skin, full pain-free stimulated paretic hand opening, and full less-affected-side hand opening. Exclusion criteria included other upper-limb-affecting neurological conditions, severe communication/cognitive impairment, cardiac arrhythmias, electronic implants, hemi-neglect, insensate upper-limb, and seizure or botulinum toxin injection within three months.

TABLE II.

Participant Demographics & Dose of CCFES Video Game Therapy Received (Excluding Rests and Setup)

ID	Sex	Age (yrs)	MACS level	Time of 1st lesion	More Affected Side	Lesion Details	Total Therapy Hours	Functional Task Practice Duration Hours	Hand Openings / Hour for Task Practice	Video Game Hours	Hand Motions / Hour for Video Games	Home Sessions Done/ Assigned (%)
P1	M	11	II	Prenatal	R	Bilateral brainstem intraventricular hemorrhage	46.3	5.6	60	40.7	1647	139/142
P2	M	9	III	8 months	R	Left hemispherectomy causing right thalamic ischemia	45.9	3.1	71	42.8	1020	161/169
P3	M	8	III	Prenatal	L	Right frontoparietal cystic encephalomalacia	33.1	5.7	45	27.4	1453	244/251

P1 and P2 were assigned 15-min game sessions, but P3 was assigned 7-minute sessions.

F. Pre-Treatment Setup and Instruction

During a setup visit, study staff determined electrode placement, stimulation parameters, and provided participants and caregivers with photographs of electrode placement, skin markings surrounding electrode placement locations, user manuals, and device instruction. Participants were instructed on electrode placement, stimulator operation, command glove and sensor mitt donning/doffing, and computer/game operation. Participants and their caregivers were then required to practice device operation in the lab until they could demonstrate competency without staff assistance. A PC identical to the lab PC was installed at the participant’s home, where study staff again reviewed device operation until the child and caregiver could demonstrate competency without staff assistance.

G. Treatment Regimen

The six-week treatment consisted of lab visits twice per week in weeks 1-3, once per week in weeks 3-6, and daily home sessions. Lab visits included up to 60 minutes of video games (four games, 15 minutes/game) and up to 45 minutes of therapist-guided functional task practice (board games, cards, puzzles, snacks, bean bags, laundry, dinnerware, and utensils). At home, participants were assigned up to ten 45-minute therapy sessions/week (three games, 15 minutes/game). Actual dose can vary because the number of games assigned for home use was determined by how long the child could tolerate the games in the lab, which gradually increased during weeks 1-3.

H. Outcome Measures

Adherence was assessed by the number of assigned home sessions completed versus the number assigned based on game usage logs. The number of hand motion repetitions was also recorded to ensure that participants were actively using their paretic hand. Staff also counted the number of hand opening attempts during functional task practice during lab sessions.

Qualitative assessments included participant diaries and end-of-treatment questionnaires. In the diaries, participant were asked to report any issues related to home use, including device malfunctions or comments about game difficulty. The end-of-treatment questionnaire asked participants to report perceptions about the intervention’s effect on hand opening/closing/use, opinions on therapy duration, ease of device/game operation, whether they wished that the intervention was part of their conventional treatment, and their favorite/least-favorite games.

Upper limb motor outcomes were assessed at baseline, treatment end, and four-week follow-up. One motor function outcome was the Assisting Hand Assessment (AHA), which measures the level of spontaneous paretic hand use during 22 bimanual activities [43]. The quality of paretic hand use during each activity is rated on a 4-point scale and the total maximum score is 100 logit points. The minimum clinically important difference (MCID) for the AHA is unknown, but the minimum detectable change (MDC) has been reported as 4 logits [43].

Another motor function outcome was a 30-second sinusoidal tracking task with amplitude varying between 20–80% of volitional index finger extension on the more-affected hand. Average error percentage was defined as the total absolute distance between the cursor and the track for each track segment, summed over the entire track and normalized by the length of the track, as defined by the following equation.

$$\mathrm{Average\; Error}\%=\frac{{\sum }_{\text{n}=0}^N\left|y_{\mathrm{cursor}}\left[\text{n}\right]-y_{\mathrm{track}}\left[\text{n}\right]\right|}{\text{N}}\times 100\%,$$

where N was the number of discrete segments making up the sinusoidal track, y_cursor and y_track were values between 0 and 1 representing the cursor and track screen positions, respectively. After one practice trial, maximum average error of three trials (with one-minute rest) was reported as a percentage of the volitional range of the paretic hand (maximum 50%). The percent average error can be interpreted as a percentage of the paretic hand’s volitional range of motion. Changes exceeding 10% are often considered important in the literature [5], [44].

Motor impairment of participant 1 (P1) and Participant 2 (P2) was assessed with the upper extremity Fugl-Meyer assessment (FM) [45] to compare against prior adult studies, though it has not been validated for children. Participant 3 (P3) was assessed with Melbourne Assessment 2 [46] (MA2), which has pediatric psychometrics. We did not assess both FM and MA2 concurrently in order to minimize assessment duration and prevent losing participant attention. Both MA2 and FM are similar in that they measure the participant’s ability to perform a series of unimanual movements and tasks. FM has a cumulative maximum score of 66 points, while MA2 rates the performance of each test item as a percentage of maximum possible score in the subscales of Range of Movement, Accuracy, Dexterity, and Fluency. Pediatric MCID for upper extremity FM is unknown, but has been reported as 4.25 points for adults with stroke [45]. Category MCIDs for MA2 are reported to be 8.7% for Range-of-Movement, 12.8% for Accuracy, 13.1 for Dexterity, and 10.6% for Fluency [46].

IV. Results

A. Participant #1 (P1, Age 11, Moderately-Severe, Manual Activity Classification System (MACS) [47] level II)

Prior to participation, P1 received one hour/week of occupational therapy (OT) and one hour/week of physical therapy (PT) at Cleveland Clinic, which were stopped until after follow-up assessments were completed. During the study, he continued to receive applied behavior analysis and participated regularly in school-based interventions including PT, speech language therapy, and peer social group. school-based OT was received less frequently on an as-needed basis.

Adherence:

P1attended all 9 lab visits and learned Paddle Ball, sound Tracker, skee Ball, and Marble Maze in that order at the first four lab visits. Each game was assigned for home use after P1 was able to tolerate 15 minutes of the game without rest. As shown in Table II, P1 completed 139 of 142 assigned home sessions (98%). The three unfinished games occurred during week 6. In total, he received 46.3 hours of actual paretic hand use. This consisted of 40.7 hours of CCFES video game therapy and 5.6 hours of functional task practice. His number of hand motion repetitions/hour was 1647 for video games and 60 for functional tasks in the lab.

Qualitative:

P1 needed assistance with electrode placement at first, but became independent in week two. In the end-of-treatment survey, he responded “no” to whether he thought the intervention improved his hand opening or closing. However, he responded “yes” to whether the intervention improved his hand use, reporting improved ability to “hold cups and phones.” He responded “yes” when asked if he thought the treatment duration was too long and he suggested a four-week duration with two half-hour home sessions. He responded “yes” when asked if he thought that the intervention should have been part of his original care. His favorite game was skee Ball and Marble Maze was his least favorite because it became boring.

Motor Outcomes (Fig. 4):

Fig. 4.

A) Fugl-Meyer Upper Extremity (assessing impairment) for P1 and P2 only, B) Melbourne Assessment 2 of Upper Limb Function for P3 only (assessing impairment), C) Sinusoidal wave finger tracking (instrumented assessment), and D) Assisting Hands Assessment (assessing function).

Upper extremity FM increased from 28 at baseline to 41 at treatment-end and 35 at follow-up. The hand FM subsection (maximum 14) increased from 3 at baseline to 9 at treatment-end and 7 at follow-up. Tracking error of the paretic hand improved from 16.1% of volitional range of motion at baseline to 6% at treatment-end and 4.4% at follow-up. AHA increased from 52 at baseline to 63 at treatment-end and 64 at follow-up.

B. Participant #2 (P2, Age 9, Severe Hand Impairment, No Volitional Hand Opening)

Prior to participation, P2 received 1 hour/week of outpatient OT and 1 hour/week of PT, which were stopped until after follow-up assessments. He maintained twice weekly school OT, which were necessary for schoolwork.

Adherence and Dose:

Video games were introduced in the same order as P1, but sound Tracker was limited to 7 minutes/game during week one to prevent fatigue until P2 could tolerate 15 minutes/game in week two. P2 attended all 9 lab visits and completed 161 of 169 assigned home sessions (95%). Seven home sessions were missed due to influenza and one due to malfunction of the USB data acquisition device. He attained 45.9 hours of total paretic hand use, which consisted of 42.8 hours of CCFES video game use and 3.1 hours of functional task practice (Table II). Hand motion repetitions/hour was 1020 for games and 71 for functional tasks.

Qualitative Reports:

Due to severe hand impairment, P2 needed assistance from their caregiver to don the CCFES glove and sensor mitt. At treatment end, he did not report improved hand opening, but reported improved paretic hand closing and ability to “hold a lightsaber for the first time” with the paretic hand. Also, he thought continued treatment would benefit him and that the therapy should have been part of his original care. He thought six weeks of therapy was just right, but suggested shorter, 30-minute home sessions. Marble Maze was P2’s favorite game and Paddle Ball was his least favorite because it was challenging.

Motor Outcomes (Fig. 4):

FM increased from 24 at baseline to 31 at treatment-end, and 29 at follow-up. The FM hand-only subsection increased from 3 at baseline to 5 at both treatment-end and follow-up. Tracking task error improved from 36.5% at baseline to 18.7% at treatment-end and 23.8% at follow-up. AHA improved from 29 at baseline to 32 at treatment-end and 28 at follow-up.

C. Participant #3 (P3, Age 8, Moderate Hand Impairment, MACS Level III)

P3 maintained 20 mg of Baclofen orally three times/day and kept using a dynamic movement orthosis during the day, but was not receiving any therapy prior to participation.

Adherence:

Video games were introduced in the same order as P1 and P2, but game durations were reduced to seven minutes because P3 had difficulty maintaining attention for 15 minutes. This required the number of game assignments to be doubled. Also, P3’s treatment ended at week five to accommodate his participation in previously-scheduled CIMT lasting 3 weeks. P3’s end-of-treatment assessment occurred prior to starting CIMT and his 4-week follow-up occurred after CIMT, which included outpatient occupational therapy five days per week, three hours per day and also required his less-affected upper extremity put in a cast the entire three weeks. He completed 244 of 251 assigned home sessions (97%), missing four games due to hand fatigue and three games were due to stimulator battery failure. In total, P3 attained 33.1 hours of paretic hand use, which consisted of 27.4 hours of CCFES video games and 5.7 hours of functional task practice (Table II). Hand motion repetitions/hr was 1453 for games and 45.4 for functional tasks.

Qualitative Reports:

P3 did not need caregiver assistance for home sessions from week 3 onward. At treatment end, he reported improved paretic hand opening, closing, and ability to “hold an iPad [using the paretic hand] for the first time.” He thought continued treatment would benefit him and that the therapy should have been part of his original care. Skee Ball was his favorite game and Sound Tracker was least favorite because it became boring (even though he did play it at home).

Motor Outcomes (Fig. 4):

AHA improved from 43 at baseline to 62 at treatment-end and 64 at follow-up (after CIMT). Average tracking task error improved from 17.7% at baseline to 7.8% at treatment-end and regressed to 20.1% at follow-up (after CIMT). MA2 Range-of-Movement subscale increased from 37% at baseline to 59% at treatment-end, and regressed to 29% at follow-up (after CIMT). Dexterity MA2 subscale increased from 48% at baseline to 76% at treatment end, and regressed to 52% at follow-up (after CIMT). Fluency MA 2 subscale increased from 29% at baseline to 43% at both treatment-end and at follow-up (after CIMT).

V. Discussion

A. Adherence and Dose

Participants achieved 4.6-7.1 hours/week of home use. Although these averages were less than the target 7.5 hrs/week, they exceeded the dose of other home-based trials in children with hemiplegia using video game consoles (2-3 hours/week) [48], internet-based tele-rehabilitation (1.6 hours/week) [49], conventional occupational therapy (1.2 hours/week)[50], and CIMT (3.9 hours/week) [51]. Only home bimanual intensive training reported a greater dose of 9.6 hours/week [52]. Hand motion repetitions/hour during games were 1020-1647, which exceeded the 640 shoulder-elbow repetitions/hour reported for children during robot-assisted video games [53]. This indicates that CCFES-assisted video games can deliver high doses of at-home therapy relative to other home-based interventions.

We discovered that factors that can limit dose and adherence include the need for children to build up stamina to play the games for extended periods of time, as they may not be used to constant use of the paretic hand. Other factors may include short attention span of the child and participation in other therapies that preclude other interventions, such as CIMT, which requires the less-affected hand to be placed in a cast during the duration of treatment in order to force paretic hand use at home. P2’s caregiver was observed to motivate the child during week 5 by promising a “special treat” if the child maintained adherence through the final week of treatment. We also heard from another therapist unrelated to the study that P3’s had to be reminded frequently to maintain adherence to the home portion of the protocol. Based on this anecdotal information, we expect adherence to be more challenging during the school year and there may need to be frequent short-term rewards for the children to facilitate adherence at home.

B. Motor Outcomes

Motor outcomes improved at treatment-end for all participants. Fugl-Meyer increased 13 and 7 points for P1 and P2, respectively, exceeding the 4.25-point minimum clinically important difference (MCID) for adults [45] (pediatric MCID is unknown). These improvements were comparable to the 8-point gain reported in a study of pediatric robot-assisted video games [53] and exceeded 2-3 point gains in our own study of three adults treated with CCFES video games [7]. For P3, three of the four MA2 subscales increased beyond clinically-important differences at treatment-end [46]: Range-of-Movement (22.3% improvement; MCID 8.7%), Accuracy (28% improvement; MCID 12.8%), and Fluency (18.3% improvement; MCID 10.6%).

All participants increased bimanual hand use at treatment-end. Assisting Hands Assessment increased 11-17 logits, exceeding the 4-logit MDC (MCID unknown) [43] and gain of 2 logits for tele-rehabilitation [49], 3 logits for at-home bimanual intensive training [52], and 2 logits for CIMT [51].

Finger tracking also improved at treatment-end by 8-17% of volitional hand opening range. There are no established clinically-important thresholds for sinusoid tracking accuracy, but changes exceeding 10% of volitional paretic hand opening range of motion are often considered important in the literature.

The majority of the motor outcome trajectories showed improvement at end of treatment and then regression at follow-up. This supports the notion that the improvements are associated with the intervention and are not within the normal variability of the outcome measure of the participants. The lack of sustained improvement at 1-month may be because the gains achieved during the treatment phase were not large enough to keep the child using their more-affected hand; therefore, they lost the gains they achieved. P3’s maintained AHA gains at follow-up could be due to the fact that he was forced to use his more-affected arm between treatment-end and follow-up assessments while participating in CIMT. Future studies may investigate whether larger and longer-lasting gains could be produced by optimizing the treatment dosage and duration. Specifically, it is important to investigate the effect of dose frequency and total duratiom. Our CCFES studies in adults with stroke found that 12 weeks of intervention had longterm benefits [3], [4], [6]. However, even if a higher dose benefits children, it may not be viable due to school and extracurriculars. It is also important to study the efficacy of CCFES apart from hand therapy video games. If CCFES alone has benefit, the intervention could be better integrated to with everyday activity (analogous to casting the less-affected hand in CIMT).

The current results are consistent with prior studies showing that temporarily linking motor intention to motor output and sensory feedback from the affected limb may facilitate neuroplastic changes leading to motor recovery [54]. These results could be associated with evidence that synchronizing the movement of both hands during bilateral arm training can lead to facilitatory cortical changes [55].

C. Limitations

The current findings may be biased by small sample size, lack of randomized assignment to a control group, outcomes assessed by an unblinded assessor, unrepeated baseline assessments, and short follow-up period. Furthermore, the participation of P3 in clinical CIMT between treatment end and follow-up is a potential confounder on the effect of the studied intervention. Study adherence may be biased by the fact that this study occurred primarily during the summer vacation for the participants. Additional studies are needed to investigate how adherence is affected by participation during the school year. Also, the existing hand therapy video games might be further developed to maximize engagement and additional games can be designed to add greater variety of skill training. Developing strategies for maximizing the child and family’s motivation to adhere to the intervention is also important.

VI. Conclusion

The key findings of the current study were that three children with moderate-to-severe hemiplegia and their caregivers were able to administer CCFES-assisted video game therapy at home to achieve a high dose of paretic hand use. All participants achieved meaningful motor outcome gains at end of treatment. Further development to maximize home adherence and additional studies with greater sample size and experimental rigor are warranted. These results have already informed the design of an ongoing controlled trial (National Clinical Trial #02925455) and are valuable for developing future home-administered interventions for children with unilateral cerebral palsy and hemiplegia.

Fig. 3.

(Left) Screen shots from Paddle Ball (upper left), Skee Ball (upper right), Sound Tracker (lower left), and Marble Maze (lower right).