Design and Usability of a Home Telerehabilitation System ...wkdurfee/publications/Track-JBiomechEng.pdfupload tracking data to a central file server. The system was designed for use

Track-JBiomechEng.doc Page 1 of 31

Design and Usability of a Home Telerehabilitation System to Train Hand Recovery Following Stroke William K. Durfee1, Samantha A. Weinstein1, James R. Carey2, Ela Bhatt2, and Ashima Nagpal2

1 Department of Mechanical Engineering, 2 Program in Physical Therapy, University of Minnesota, Minneapolis USA Submitted to the Journal of Biomechanical Engineering (ASME), April 2006 Address Correspondence to: William Durfee University of Minnesota 111 Church ST S.E. Minneapolis, MN 55455 tel: 612-625-0099 email: [email protected]


ABSTRACT

Current theories of stroke rehabilitation point towards paradigms of intense, concentrated use

of the afflicted limb as a means for motor program reorganization and partial function

restoration. TrackTrain is a home system for stroke rehabilitation that trains recovery of hand

function by a treatment of concentrated movement. A wearable goniometer measured finger

and wrist motions in both hands. An interface box transmitted sensor measurements in real-

time to a laptop computer. Stroke patients used joint motion to control the screen cursor in a

one-dimensional tracking task for several hours a day over the course of 10 to 14 days to

complete a treatment of 1800 tracking trials. A tele-monitoring component enabled a therapist

to check in with the patient by video phone to monitor progress, motivate the patient and

upload tracking data to a central file server. The system was designed for use at home by

patients with no computer skills. TrackTrain was placed in the homes of 24 subjects with

chronic stroke and impaired finger motion, ranging from two to 305 miles away from the

clinic, plus one that was a distance of 1,057 miles. Fifteen subjects installed the system at

home themselves after instruction in the clinic, nine required a home visit to install. Three

required follow-up visits to fix equipment. A post treatment telephone survey was conducted

to assess ease of use and most responded that the system was easy to use. Functional

improvements were seen in the subjects enrolled in the formal treatment study, although the

treatment period was to short to trigger cortical reorganization. We conclude that TrackTrain

is feasible for home use and that tracking training has promise as a treatment paradigm.


1 INTRODUCTION

Stroke is a major health problem. According to the American Heart Association (2005), in the

United States, approximately 700,000 people suffer a first or recurrent stroke each year and

stroke is the third leading cause of death after heart disease and cancer. On average, every 45

seconds, someone in the U.S. has a stroke. Stroke is the leading cause of serious, long-term

disability. About 4.7 million stroke survivors are alive today, and this number is increasing

because of improved acute care. In 1999, more than 1.1 million adults reported difficulty with

functional limitations resulting from stroke. The direct and indirect cost of stroke in 2005 was

$56.8 billion.

A common stroke disability is partial or complete paralysis of a limb, which leads to

impairment of basic functional skills in 75% of people with stroke (Jorgensen et al., 1999).

One of the most debilitating limb impairment is paralysis of the hand (Hummelsheim et al.,

1997; Nudo and Friel, 1999). Often, the individual will be able to close the fingers into a fist,

which is part of the characteristic spastic flexion synergy, but is unable to open the fingers

because of extremely weak finger extensor muscles and spastic tone in the antagonistic flexor

muscles. As a result, the person cannot manipulate objects dexterously, which impairs

feeding, dressing, handwriting, and occupational skills.

Recent scientific findings fundamental to motor learning and neuroplasticity have challenged

the traditional paradigms for motor rehabilitation following stroke. Functional loss in stroke

occurs not only from cells that have died but also from reduced excitability in neurons that

have survived the infarct (Feeney and Baron, 1986; Chu et al., 2002). The significance for

rehabilitation is that there may be viable neurons available for recruitment in the stroke


hemisphere, leading to the search for potent mechanisms that promote the restoration of

voluntary recruitment of suppressed neurons. The literature is clear that movement retraining

for stroke rehabilitation must be intense and must be prolonged and the patient must engage

the task repeatedly and independently (Schmidt, 1988, Winstein et al., 1994). Importantly,

self-improvement does not stem directly from the therapist coaching and guiding the activity;

rather, improvement is thought to stem from internal information processing by the subject

with consolidation of all the cues inherent to the task leading to a refined motor output

(Winstein et al., 1994). Thus, subjects must be given opportunities to train themselves. The

guiding and coaching efforts by therapists may be helpful in the short run but in the long run

it will be the subject’s own information processing, including failed efforts, that will lead to

the cellular and molecular changes that stimulate motor learning/relearning and a fuller

recovery.

One example of modern therapy that is based on motor learning principles is constraint-

induced movement therapy. CIMT forces the patient to use the paretic limb while the healthy

limb is restrained and was designed to overcome learned nonuse of the impaired limb (Taub,

1980). Clinical studies have shown CIMT to be effective for patients who start with some

residual range-of-motion (Liepert et al., 2000; Page et al., 2004; Ploughman and Corbett,

2004, Taub et al., 1993; Taub et al., 1999). CIMT requires massed practice (practice time is

greater than rest time) for 6-7 hours daily for two weeks with continual therapist supervision

(Taub, 2000).

Forced-use such as CIMT, however, may not be the most potent behavioral stimulus for

neuroplastic change. Forced-learning may be even more effective. This is a pivotal question

in rehabilitation as it dictates the type of tasks and level of cognitive engagement that


therapists select for training in subjects with stroke. Repetitive execution of simple motor

tasks may not be as effective in re-training motor skills as complex motor tasks that involve

in-depth cognitive processing (Carey et al., 2005). A joint-movement tracking training system

to promote recovery of movement control in stroke was applied to the hand of subjects with

stroke (Carey et al., 2002) and to the ankle in a single-subject research design (Carey et al.,

2004) with favorable results. Unlike CIMT, tracking training emphasizes motor learning

principles espoused by Schmidt (1991) that minimize guidance and involvement by a

therapist, thereby potentially saving treatment cost.

Conventional physical therapy for rehabilitation requires extensive interaction with a

therapist, often for hours a day. These labor intensive efforts are costly and contribute to the

staggering financial impact of stroke. Consequently, practitioners and administrators are

challenged to find alternative methods to deliver quality health care at reasonable cost. One

example of cost containment in stroke rehabilitation is home-based rehabilitation. Evidence

has shown that early discharge from acute hospital care followed by home-based

rehabilitation, involving professionals visiting the home, reduced costs while preserving

treatment outcomes that were equivalent to conventional rehabilitation methods (Young and

Forster, 1992, Widen Holmqvist et al., 1998, Beech et al., 1999, Anderson et al., 2000). With

home-based treatment, patients can receive treatment at an intensity (hours per day) that is

appropriate to their state of recovery rather than directed by health insurance policy, which

may be too much early on or too little later on. Conventionally, rehabilitation for people with

stroke is restricted to the first 6-12 months following the stroke, however, current research

indicates that further recovery is possible with rehabilitation years after stroke (Kraft et al.,

1992, Kunkel et al., 1999, Liepert et al., 2000a, Carey et al., 2002, Luft et al., 2004). With


home-based treatment, the amount of time devoted to rehabilitation is only restricted by the

patient’s level of commitment and not by the cost of a therapist.

Interest in technology driven, home-based stroke movement therapies is evidenced by the

emergence of pilot projects such as JavaTherapy, a web application that uses standard input

devices such as force controlled joysticks (Reinkensmeyer et al., 2002), and UniTherapy, a

telerehabilitation platform for computer-assisted movitvating rehabilitation that supports force

feedback training with standard and custom PC input devices (Feng & Winters, 2005).

Robot-aided neurorehabilitation is another emerging technology based treatment for stroke.

Grounded or wearable powered manipulators assist or guide the paralyzed limb through a

desired motion in much the same manner as a classic hands-on physical therapy treatment.

Upper limb devices include the MIT Manus (Krebs et al., 1998; Fasoli et al., 2003; Fasoli et

al., 2004), Ketic Muscles’ Hand Mentor (Koeneman et al., 2004). Lower limb devices include

the Locomat by Hocoma (Colombo et al., 2000) among many others. Preliminary results have

shown that upper and lower limb robot therapy can be effective. Unlike the TrackTrain

system described here, robots can be expensive and must be carefully designed to be safe

before being considered for home use.

Our group conducted a pilot study that used a finger movement tracking program where

subjects with stroke received 18-20 one-hour rehabilitation training sessions using their

paretic index finger to track target waveforms displayed on a computer screen under variable

conditions (Carey et al., 2002). The subjects came to the clinic for treatment sessions with a

physical therapist supervising. Pre and post training treatment functional magnetic resonance

imaging (fMRI) was used to determine which parts of the brain were activated when the

subject was performing the finger tracking activity. The results showed improved ability to


grasp and release small objects, and a change in cortical activation from predominantly

ipsilateral control (non-stroke hemisphere active) to contralateral control (stroke hemisphere

active) during movement of the paretic finger. The difficulty with the pilot study was that

subjects were required to come to the clinic for training sessions. This was not only

inconvenient, but physically difficult and time demanding of the subjects. Further, in-clinic

training sessions could last no more than 60 minutes while current massed practice

rehabilitation theory points to the benefit of longer training sessions.

In light of the promising results of the pilot study, our group has developed TrackTrain, a

system that enables tracking training sessions in the subjects’ homes without a therapist

present. With home-based training the subject only has to come to the clinic for pre and post-

treatment evaluations. TrackTrain attracted more subjects to the study than clinic-based

training because many eligible subjects are not interested in coming to the clinic daily, or live

too far from the clinic for convenient daily travel. In addition, if home-based therapy proves

to be effective than it could lead to new clinical treatment paradigms to serve larger numbers

of patients with stroke at a lower cost.

The main requirements driving the design of TrackTrain were first to create a system that

required the patient to move his or her hand and wrist in a tracking task that required

concentration to accomplish motor relearning, and second to create an easy-to-use system that

could be operated autonomously at home without requiring technical expertise. In addition,

the system needed a telecommunication mechanism for periodic real time interaction with a

remote physical therapist to monitor, assist, and motivate the patient’s progress. The basics of

the tracking task were established in the pilot study. Thus the challenge in this project was to

design system that allowed patients to train at home without a therapist present.


2 SYSTEM DESCRIPTION

2.1 Overview

The TrackTrain home tracking system has two custom bilateral electrogoniometers that sense

wrist and finger flexion and extension motions, a microcontroller-based interface box, a

laptop computer with tracking software, a cell phone, a web cam, and a land line telephone

switch box (Figure 1). The system is designed for self installation in the patient’s home, or for

installation by the therapist. While the system has many individual components, it is still

relatively compact and fits easily on a desktop or table without being an overly obtrusive

presence in the patient’s home.

<Figure 1 about here>

2.2 Hand Sensors

The hand sensor has three platforms connected by rotating links, all made of polycarbonate.

The sensor measures flexion and extension of the wrist and first MCP joints (Figure 2). The

links form two four-bar mechanisms that transfer finger and wrist joint motion to two 0.5

inch, 1.0K potentiometers (ETI Systems, SP12S-1K) located on the platforms. This design

was chosen over potentiometers axially aligned with the joint because it afforded higher

tolerance to placement location. Because the linkage straddles the joint, the two platforms at

either end need only be on opposite sides of the joint and do not require precise placement.

This is particularly important for a sensor that was designed to be self-donned by the patient.

The links are designed to rotate relative to one another in the plane of the platforms which

allows the sensor to fit multiple hands as the angle between the axis of the wrist and the axis

of the forefinger differs between people. The sensor dimensions were determined by sampling


male and female hands and consulting anthropometric data (Tilley, 2001). The relation

between the joint angle and potentiometer angle is sufficiently linear over the typical range of

motions seen that a linearization algorithm was not incorporated into the software. The sensor

had small green LEDs on the two proximal platforms which flashed as a cue to the subject for

which joint was active.


The patient must don and doff the sensor independently. The platforms balance on top of the

arm with relative stability. Hence, most of the difficulty in this task is associated with

fastening and adjusting the straps. Two slap bracelets (fabric covered, metal strips that lie flat

when bent in one direction and self-assemble into a circle when a small amount of pressure is

applied in the opposite direction) were used for anchoring the proximal sensor platform to the

forearm. A Velcro strap with a D-ring at one end was used to anchor the hand platform. The

patient slides his hand into the looped strap and then pulls it back around on itself to tighten

and secure the Velcro. The finger strap is flexible fabric ring that the patient slides his finger

into. A spring loaded clip is used to pull up the slack in the ring and secure it at the correct

diameter. Because the clip is difficult to operate with limited dexterity and strength, the ring

size is adjusted by the therapist at the initial fitting session. Because the finger is conical, the

patient can don the sensor by sliding his forefinger into the ring until it is snug. The links can

be used as handles for device removal.

2.3 Interface box

The sensor interface box converts the potentiometer reading into a digital signal that is

transmitted to the computer through its serial port. The interface box contains analog signal


conditioning circuitry including programmable gain and offset, a 20 Hz, two-pole low pass

filter, and a programmable multiplexer that selects which channel to digitize (Figure 3).

Signals are digitized at 100 Hz with a 10-bit analog-to-digital converter. Software on a

PIC16F873 microcontroller controls the peripherals, creates the real-time sampling loop, and

sends the signal to the PC serial port. The host program running on the PC communicates

with the microcontroller to enable and inhibit sampling, and to set current channel, gain and

offset. A four stage running average filter implemented on the PC performed additional

smoothing of the sampled signal.


The sensor interface box has a large red button on the top cover, a small black button on the

back, and two color coded jacks for the left and right hand sensor cables on the front . The red

button is the only control the user needs to operate the program. Through prompts on the

computer display, the red button is used to start the treatment, pause the treatment, calibrate

the hand sensors, and end the treatment. The black button is used for unusual circumstances to

abort the program and immediately shutdown the computer.

2.4 Computer

The system control unit is a Dell Latitude D600 laptop computer running the Windows XP.

The computer is treated as an appliance with the patient never having to use the keyboard or

mouse. The laptop mouse was not connected and the keyboard was covered with a screen.

The keyboard and mouse were not used for several reasons. Keys are close together which

increases the likelihood that a patient with limited dexterity might press the wrong key. Using

a mouse requires fine motor control which is lacking in the users, and also requires familiarity


with computers. Access to the keyboard and mouse could afford technologically savvy

patients a means of getting into the inner workings of the system, which is not desired. The

red button on the interface box and the laptop’s power button were the only two controls

needed by the subject to operate the system. On power-up the computer immediately boots

into the tracking program and from there, all interaction by the user was done with the single

red button on the interface box.

2.5 Software

The tracking application on the PC was created using Microsoft Visual Basic 6.0. On power-

up, the PC automatically started the application. With both sensors mounted, the subject went

through a prompted calibration sequence with the display instructing the subject to open and

close their hand and flex and extend their wrist as far as possible which automatically

communicated the subject’s active range of motion to the application (Figure 4). The red

button was pressed to sequence through the joints. If the measured range of motion was less

than a threshold, it was likely that the sensor had been misapplied and the subject was advised

through on-screen instructions to telephone the therapist for advice. The sensors must be

calibrated each time the program is started as there is no guarantee that the sensor was placed

on the hand the same way each time. At the completion of calibration, the subject pushed the

red button to commence the day’s therapeutic tracking training session.


The first screen of a trial contained four key pieces of information for the subject (Figure 4).

At the top of the screen, the number of blocks left in the session and the number of trials left

in the block are shown. Instructions on how to pause the program by pressing the red button


are given in the center of the screen. Text and pictorial indicators of the hand, joint, and arm

position to be used for the next tracking task are shown at the bottom of the screen. There

were two LEDs on the hand sensors, one located at the wrist and the other at the finger. The

LED on the hand and joint to be used in the next trial blinked throughout the first and second

trial screens indicating which joint would be used for the next tracking trial. A countdown

clock displayed the time remaining to the start of the next tracking task.

The second screen in the trial sequence was the tracking task (Figure 5). At the start of the

trial, a green ball was displayed on the far left of the box with the joint motion controlling the

vertical motion of the ball. The patient had one second at the start of the trial to move the ball

up and down to get a feel for the range of motion through which he would be working. After

the one second delay, a beep sounded and the ball began to move horizontally across the

screen at a fixed rate determined by the waveform duration. The task of the subject was to

move the green ball follow the blue target by moving the specified joint up and down. The

ball left a red trail for visual feedback to the patient. When the ball reached the right end of

the box, a double beep indicated the end of the trial. The third trial screen showed a total

tracking score to indicate the subject's performance (Figure 5).


Eleven tracking performance scores (Carey, 1990; Schmidt and Lee, 2005) were computed

and stored for each trial. The scores and associated computation algorithms are shown in

Table 1. The subject is shown 1000*AI during the rest period after each trial as their

performance score on the trial. The score ranges from a large negative number if the tracking

is anti-phase to 1000 for perfect tracking.


<Table 1 about here>

Following some trials, the screen also displayed text providing automated feedback on

performance. The feedback alerted the subject to areas of weak performance, such as wrist or

finger extension, or alerted the subject to excessive or insufficient frequency of movement.

The feedback was generated based on the 11 tracking performance metrics that were

calculated after every trial. This more descriptive information is known as “knowledge of

performance” (KP). KP feedback was shown to subjects in a faded schedule after certain

blocks of trials being shown every third block on day one and gradually declining to every

20th block by day 10. The purpose of the faded feedback is to reduce reliance on external

guidance and instead promote internal error-detection capability (Schmidt, 1991). The KP

information, when it appeared, was based only on the immediately preceding trial and was

always complemented with an encouraging comment to motivate continuing effort. An

example is, "You are not reaching the tops of the upper blue waves. Try to reach higher and

touch the peaks," and "Mrs. Jones, your performance is great! Keep it up!"

The treatment plan was 180 tracking trials per day for 10 days, a total treatment of 1800 trials.

The trials were organized into 60 blocks with three identical trials per block. The blocks were

randomly selected from a set of 100 test conditions which varied in target wave shape,

frequency, amplitude, duration, joint used, hand used, hand posture and visual feedback

(Table 2). The variations were needed to keep the patient concentrated on the task. The paretic

hand was used for 90% of the blocks and the nonparetic hand for 10%. Tracking with the

nonparetic hand was included to take advantage of cross education. Cross education is the

principle of training in which effort unilaterally has beneficial effects bilaterally (Greenough


et al., 1985; Imamizu et al., 1998; Shields et al., 1999). Hand posture (palm up/down), where

for the up case, cursor movement is opposite the joint movement, was included to case greater

complexity and depth of cognitive processing for the task (Carey et al., 1994, 1995, 1998). In

some blocks, the ball and red trail were replaced by a vertical hairline cursor that sweeps

across to show timing but provides no visual feedback on joint location. Every trial was

followed by a rest period of the same duration of the preceding trial, constituting distributed,

rather than massed, practice (Dail, 2004; Lee, 1989).


Once the training session started, other than tracking, the only action required by the subject

was to press the red button whenever a pause was desired. If the subject did not press the

pause, the computer continued until all 180 trials were done that day. Most subjects choose to

pause periodically. If the pause was less than five minutes, pressing the red button to resume

brought the subject immediately back to the previous training point. For longer pauses, the

program assumed the subject had removed the sensors and pressing the red button to resume

went back to the calibration screen. The application informed the subject when the day’s

session was complete and automatically shut down the computer. The automatic shutdown

regulated the rate of rehabilitation to the prescribed regimen of one session per day. The

application had additional branch points to detect and exit gracefully if the sensor was still

(meaning the subject may have left the room), the sensor was disconnected from the interface

box, the computer was shut down, or power was removed from the sensor box.

2.6 Telecommunication


Telecommunication took place between therapist and patient every few days. The periodic,

real-time, face-to-face interaction with a therapist was for monitoring, motivating and

assistance, and is essential for patient compliance with the rigorous protocol. We assumed

every subject had a land-line telephone and cellular phone coverage, but not broad-band

internet. Based on pilot work, we assumed that high quality audio was essential for

meaningful communication, but low quality video was adequate. Voice communication was

by cell phone (TracPhone). Two-way video and data transmission was by web cameras

(Logitech QuickCam 4000) and dial-up internet access, resulting in low quality (color,

128x96, 3 frames per sec) video imaging.

To initiate a tele-session, the therapist called the subject on the cell phone (after first alerting

the subject with a landline phone call as the subject is not used to answering the cell phone).

The cell phone was set for auto-answer on any button push so that motor-impaired subjects

did not have to manipulate small buttons. A cell phone speaker phone accessory was used so

that the subject could communicate hands free. The therapist instructed the subject to flip the

switch on the telephone interface box, which connected the land line to the PC modem, then

power up the computer. On boot up, the PC automatically detected the dial tone, dialed the

ISP, connected to the host computer, and from there the therapist could download trial data

and remotely start the web cam. After connecting, the subject could see the therapist’s face

on the computer display, and the therapist could see the subject or could ask the subject to

hold his or her hand in front of the camera to see whether the electrogoniometer was attached

properly. During the communication the therapist answered questions and provided helpful

comments and motivation. At the end, the therapist could remotely shut down the subject's

computer. The session ended with the therapist instructing the subject to toggle the switch


setting on the telephone interface box so that the next time the subject computer powered up,

it would boot into tracking treatment mode.

3 EVALUATION STUDY

3.1 OBJECTIVES

The tracking training evaluation study had two objectives. First, to determine if home-based,

self-administered tracking training with only occasional teleconferencing with a remote

therapist is feasible, and second whether tracking training was more beneficial for cortical

reorganization and improved motor control than movement training. While both questions

were examined in the study, this paper is concerned with the first objective. A companion

paper (Carey et al., 2006) contains the complete results for the second objective.

3.2 METHODS

The tracking training system was placed in the homes of 24 subjects ranging from 2 to 307

miles from the University (median = 18 miles), plus one 1057 miles away. Subject inclusion

criteria were post-stroke duration of at least 12 months; satisfactory corrected vision to detect

the position of a cursor on the computer screen as being slightly above, below or exactly on a

target line; and at least 10 degrees of active extension-flexion movement at the index finger

metacarpophalangeal (MP) joint of the paretic hand. All subjects had sufficient cognition to

perform the training and testing with Mini-Mental State Examination (Folstein, 1975) scores

of 25 or higher except for one subject with a score of 21, who was still deemed appropriate to

participate. Exclusion criteria were indwelling metals or medical implants incompatible with

functional magnetic resonance imaging (fMRI), pregnancy and claustrophobia. The mean age

(± SD) of the subjects was 66.7 ± 9.6 years. The mean stroke duration (time from stroke onset


to entrance into study) was 39.1 ± 24.8 months. The study was approved by the University of

Minnesota IRB.

Twenty of the 24 subjects were in the controlled training study with 10 in the tracking training

group and 10 in the movement group. Subjects in the movement group performed the same

number of movement trials, as directed by the computer display, but neither target waveform

nor response line was displayed and thus the subject was not required to problem-solve nor

concentrate to perform the task.

During the pre-treatment evaluation session at the University, subjects were trained in use of

the equipment, and watched a video clip that showed the steps for setting up the equipment at

home. The video was stored on the tracking laptop for later review at home. The subject and

family member practiced the equipment setup and performed six tracking trials under

supervision of the therapist. The subject was then sent home with the equipment. Subjects

completed the 1800 trials in 10 to 14 days. The therapist communicated with the subject about

5 times using the teleconferencing system. The therapist was also available by pager.

Pre and post treatment evaluations included Box and Block (Mathiowetz, 1985), Jebsen-

Taylor Hand Function (Jebsen, 1969), finger range of motion, finger movement tracking

accuracy, and fMRI voxel count, intensity and laterality index. The fMRI measures were

included to evaluate cortical reorganization. Details of these evaluations are in Carey et al.

(2006). A telephone usability survey was conducted following the completion of the

treatment. The survey contained six questions with a five point Likert response scale.


3.3 USABILITY RESULTS

Fifteen of the 24 subjects were able to install the system by themselves while nine required a

home visit for installation. An additional four home visits were required to correct a computer

or equipment malfunction or equipment. One subject dropped out because they were unable to

independently don and doff the sensor and had no companion to help. Teleconferencing

sessions were conducted with 22 of 24 subjects. All 20 subjects in the controlled study

completed the 1800 trial treatment. Nineteen of the 24 subjects completed the usability survey

whose results are shown in Table 3. Two of the 19 had no teleconferencing sessions and did

not answer Question 6.


Off those subjects completing the usability survey, all agreed they "understood how to use the

system" and that "the computer was easy to use." Fourteen could independently don and doff

the sensors. Fifteen did not feel the system interfered with their daily routine, 15 saw the

automated feedback as being useful and 16 found the teleconferencing sessions easy to

manage. In self reported comments associated with the survey, almost all of the subjects

commented on the convenience of being able to work on the treatment at their leisure, fitting

it around their normal daily routine. About one quarter of the subjects had difficulty donning

and doffing the sensors by themselves, which was probably correlated to severity of the

stroke. The most consistent complaint was putting on the electrogoniometers independently,

particularly using the paretic hand to apply the electrogoniometer to the nonparetic hand,

which took some subjects 10-15 minutes or more.


3.4 TREATMENT RESULTS

Complete information on treatment results are in the companion paper (Carey et al., 2006)

and only a summary is presented here. The Track group showed improvement in Box and

Block, Jebsen Taylor, finger range of motion and finger tracking while the Move group

showed improvements only in Box and Block and Jebsen Taylor. Unlike our earlier clinic

based finger tracking training project that demonstrated brain reorganization (Carey, 2002),

the fMRI data in this study showed no significant evidence of brain reorganization for either

group.

4 DISCUSSION

The subjects were tolerant of the TrackTrain system. The concern that elderly people

unaccustomed to complications of electronic technology might drop out of the study did not

materialize. The subjects simply needed to be gently trained in the system operation and be

reassured that help was only a phone call away. Subjects who were hesitant at first became

excited by the challenge to continually do better. The evaluation project confirmed that an

easy-to-use, simple user interface, coupled with an initial instruction session is essential for

successful home use.

The hand sensor could be improved to make it easier to don and doff. Next generation sensors

should be wireless to eliminate cable interference, should not have the linkage bars above the

wrist, and should have a simplified attachment process. Because of the theoretical value of

cross-education (Greenough, 1985; Imamizu , 1998; Shields, 1999; Luft, 2004), the training

effort with the nonparetic hand should be retained, increasing the challenge for a usable

sensor design.


The current system takes up most of a desk when all the cables, boxes and communications

equipment is placed. Future systems must be compact and even simpler to set up. In

particular, the number of connections must be reduced by either combining system

components into a single appliance, or mounting components on a single board or in a single

package. Ideally, the subject should only have to plug into a power outlet and connect to the

telephone or broadband.

Periodic therapist contact was essential for reassuring and motivating patients to complete the

protocol. Visual contact by video proved important for two reasons. First, the visual presence

of the therapist often provided sufficient motivation for the patient to continue the treatment.

Second, if the patient was having difficulties with the sensor, the therapist could see the

problem and provide a solution. Future systems would benefit from a pan-tilt-zoom camera

that could be remotely controlled by the therapist. The teleconferencing technology used in

this project demonstrated that while high quality audio is essential, low quality video is

acceptable. Broadband connections are preferable to increase the quality of the video and

simplify the process of automatically downloading tracking data to a central database each

night. While the penetration of cable and DSL broadband to remote areas is still lagging, the

recent rise of wireless broadband access through cellular phone networks promises broadband

coverage to most areas.

The controlled study showed that tracking training had a positive effect on finger range of

motion, finger tracking, and functional hand dexterity, demonstrating the potential for clinical

efficacy. We suspect that the two weeks of training was too short to differentiate between the

Track and Move groups and too short to show the brain reorganization that was evident in our

earlier study. Future studies will spread the treatment over a longer time period.


One might argue that training by tracking waveforms on the screen is boring and instead that

joint motion should drive a video game. There are at least three reasons for not taking the

video game approach. First, it is not clear what video game would be motivating for hours at a

time to elderly subjects with stroke. While teens can become addicted to a single video game

or to a category of video games, motivation drivers have not been studied for our target

population. Thus, choosing any one or any collection of video games may not solve the

motivation problem. Second, controlling a video game does not permit control over

amplitude, frequency and type of movement. With structured waveforms, TrackTrain can

provide targets at several frequencies and specific amplitudes scaled to the patient’s active

range of motion, and vary parameters in predetermined or adaptive paradigms as needed for

treatment. Third, in this study patients were sufficiently motivated by target tracking to

complete long, daily sessions of treatment.

5 CONCLUSION

The study demonstrated the technical feasibility of a home based tracking training system for

motor rehabilitation following stroke. Home-based treatment using TrackTrain resulted in

functional improvements similar to those seen for the equivalent clinic-based system and

subjects indicated that they preferred the home system for its convenience. The ability of

subjects to successfully complete a treatment plan showed that elderly patients can understand

and use the system without continual supervision intended. Improvements are needed to

eliminate the sensor wires and to simplify telecommunication. Future treatment must be

spread out for longer than two weeks for significant cortical reorganization.


ACKNOWLEDGMENTS

This work was supported by the National Institute on Disability and Rehabilitation Research

(U.S. Department of Education #H133G020145-04) and the National Institutes of Health

(National Center for Research Resources #M01-RR00400 and Biomedical Technology

Research Resources #P41-R008079).


REFERENCES AHA (2005) Heart and Stroke Statistics, 2005 Update. American Heart Association. Anderson C, Rubenach S, Mhurchu CN, Clark M, Spencer C, Winsor A (2000) Home or

hospital for stroke rehabilitation? results of a randomized controlled trial : I: health outcomes at 6 months. Stroke 31:1024-1031.

Beech R, Rudd AG, Tilling K, Wolfe CD (1999) Economic consequences of early inpatient discharge to community-based rehabilitation for stroke in an inner-London teaching hospital. Stroke 30:729-735.

Black JE, Isaacs KR, Anderson BJ, Alcantara AA, Greenough WT (1990) Learning causes synaptogenesis, whereas motor activity causes angiogenesis, in cerebellar cortex of adult rats. Proceedings of the National Academy of Sciences of the United States of America 87:5568-5572.

Bohannon R, Smith M (1987) Interrater reliability of a modified Ashworth scale of muscle spasticity. Phys Ther 67:206.

Brennan DM, Georgeadis AC, Baron CR, Barker LM (2004) The effect of videoconference-based telerehabilitation on story retelling performance by brain-injured subjects and its implications for remote speech-language therapy. Telemedicine Journal & E-Health 10:147-154.

Carey J, Baxter T, Di Fabio R (1998) Tracking control in the nonparetic hand of subjects with stroke. Arch Phys Med Rehabil 79:435-441.

Carey JR (1990) Manual Stretch: Effect on Finger Movement Control and Force Control in Stroke Subjects with Spastic Extrinsic Finger Flexor Muscles. Arch Phys Med Rehabil 71:888-894.

Carey JR, Patterson RP, Hollenstein PJ (1988) Sensitivity and reliability of force tracking and joint-movement tracking scores in healthy subjects. Phys Ther 68:1087-1091.

Carey JR, Bhatt E, Nagpal A (2005) Neuroplasticity Promoted by Task Complexity. Exercise and Sport Science Review 33:24-31.

Carey JR, Anderson KM, Kimberley TJ, Lewis SM, Auerbach EJ, Ugurbil K (2004) fMRI analysis of ankle movement tracking training in subject with stroke. Exp Brain Res 154:281-290.

Carey JR, Kimberley TJ, Lewis SM, Auerbach E, Dorsey L, Rundquist P, Ugurbil K (2002) Analysis of fMRI and Finger Tracking Training in Subjects with Chronic Stroke. Brain 125:773-788.

Carey J, Durfee W, Bhatt E, Nagpal A, Weinstein S, Anderson K, Lewis S (2006), Tracking vs. movement Telerehabilitation training to change hand function and brain reorganization in stroke, Submitted to Neurorehabilitation and Neural Repair.

Cauraugh J, Light K, Kim S, Thigpen M, Behrman A (2000) Chronic motor dysfunction after stroke: recovering wrist and finger extension by electromyography-triggered neuromuscular stimulation. Stroke 31:1360-1364.

Chae J, Yang G, Park BK, Labatia I (2002) Muscle weakness and cocontraction in upper limb hemiparesis: relationship to motor impairment and physical disability. Neurorehabilitation & Neural Repair 16:241-248.

Chauhan N, Zhao Z, Barber P, Buchan A (2003) Lessons in experimental ischemia for clinical stroke medicine. Curr Opin Neurology 16:65-71.

Chen R, Yung D, Li JY (2003) Organization of ipsilateral excitatory and inhibitory pathways in the human motor cortex. Journal of Neurophysiology 89:1256-1264.

Chu WJ, Mason GF, Pan JW, Hetherington HP, Liu HG, San Pedro EC, Mountz JM (2002) Regional cerebral blood flow and magnetic resonance spectroscopic imaging findings in diaschisis from stroke. Stroke 33:1243-1248.

Colombo G, Joerg M, Schreier R, Dietz V. (2000) Treadmill training of paraplegic patients using a robotic orthosis. Journal of Rehabilitation Research and Development, 37(6):693-700.


Cramer S, Nelles G, Benson R, Kaplan J, Parker R, Kwong K, Kennedy D, Finklestein S, Rosen B (1997) A functional MRI study of subjects recovered from hemiparetic stroke. Stroke 28:2518-2527.

Dail TK, Christina RW (2004), Distribution of practice and metacognition in learning and long-term retention of a discrete motor task. Research Quarterly for Exercise & Sport, 75(2):148-155.

Desrosiers J, Bravo G, Hebert R, Dutil E, Mercier L (1994) Validation of the Box and Block Test as a measure of dexterity of elderly people: reliability, validity, and norms studies. Archives of Physical Medicine & Rehabilitation 75:751-755.

Durfee, W., Weinstein S, Carey J, Bhatt E, Nagpal A (2005), Home stroke telerehabilitation system to train recovery of hand function. Proceedings of the 2005 IEEE 9th International Conference on Rehabilitation Robotics, pp. 353-356.

S. Fasoli, H. I. Krebs, J. Stein, W. R. Frontera, and N. Hogan, (2003) Effects of robotic therapy on motor impairment and recovery in chronic stroke, Archives of Physical Medicine and Rehabilitation, 84:477-482

S. Fasoli, H. I. Krebs, J. Stein, W. R. Frontera, R. Hughes, and N.Hogan, (2004) Robotic therapy for chronic motor impairments after stroke: follow-up results, Archives of Physical Medicine and Rehabilitation, 85:1106-1111.

Feeney DM, Baron JC (1986) Diaschisis. Stroke 17:817-830. Feigin V, Lawes C, Bennett D, Anderson C (2003) Stroke epidemiology: a review of

population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Neurology 2:43-53.

Feng, X, Winters, J, (2005) UniTherapy: A computer-assisted motivating neurorehabilitation platform for teleassessment and remote therapy, Proceedings of the 2005 IEEE 9th International Conference of Rehabilitation Robotics, 349-352.

Folstein M, Folstein S, McHugh P (1975) "Mini-mental state:" A practical method for grading the cognitive state of patients for the clinician. J Psychiat Res 12:189-198.

Gomez-Pinilla F, So V, Kesslak JP (1998) Spatial learning and physical activity contribute to the induction of fibroblast growth factor: neural substrates for increased cognition associated with exercise. Neuroscience 85:53-61.

Greenough W, Larson J, Withers G (1985) Effects of unilateral and bilateral training in a reaching task on dendritic branching of neurons in the rat motor-sensory forelimb cortex. Behavioral and Neural Biology 44:301-314.

ICORR (2005), Proceedings of the 2005 IEEE 9th International Conference on Rehabilitation Robotics, June, 2005, Chicago, IL, USA.

Imamizu H, Uno Y, Kawato M (1998) Adaptive internal model of intrinsic kinematics involved in learning an aiming task. J Exp Psychol Hum Percept Perform 24:812-829.

Jebsen RH, Taylor N, Trieschmann RB, Trotter MJ, Howard LA (1969) An objective and standardized test of hand function. Arch Phys Med Rehabil 50:311-319.

Jorgensen HS, Nakayama H, Raaschou HO, Olsen TS (1999) Stroke. Neurologic and functional recovery the Copenhagen Stroke Study.887-906.

Kane RL, Chen Q, Finch M, Blewett L, Burns R, Moskowitz M (2000) The optimal outcomes of post-hospital care under medicare. Health Services Research 35:615-661.

Kimberley TJ, Lewis SM, Auerbach EJ, Dorsey LL, Lojovich JM, Carey JR (2004) Electrical Stimulation driving functional improvements and cortical changes in subjects with stroke. Exp Brain Res 154:450-460.

Kleim JA, Barbay S, Cooper NR, Hogg TM, Reidel CN, Remple MS, Nudo RJ (2002) Motor learning-dependent synaptogenesis is localized to functionally reorganized motor cortex. Neurobiology of Learning and Memory 77:63-77.

Klintsova AY, Dickson E, Yoshida R, Greenough WT (2004) Altered expression of BDNF and its high-affinity receptor TrkB in response to complex motor learning and moderate exercise. Brain Research 1028:92-104.


Koeneman, E.J., R.S. Schultz, S.L. Wolf, D.E. Herring, J.B. Koeneman. (2004) A Pneumatic Muscle Hand Therapy Device. in Proceedings of the 26th Annual International Conference of the IEEE EMBS.

Kopp B, Kunkel A, Flor H, Platz T, Rose U, Mauritz KH, Gresser K, McCulloch KL, Taub E (1997) The Arm Motor Ability Test: reliability, validity, and sensitivity to change of an instrument for assessing disabilities in activities of daily living. Archives of Physical Medicine & Rehabilitation 78:615-620.

Kramer AM, Kowalsky JC, Lin M, Grigsby J, Hughes R, Steiner JF (2000) Outcome and utilization differences for older persons with stroke in HMO and fee-for-service systems.[see comment]. Journal of the American Geriatrics Society 48:726-734.

Krebs, H.I., N. Hogan, M.L. Aisen, and B.T. Volpe, (1998)Robot Aided Neuro-rehabilitation, IEEE Transactions on Rehabilitation Engineering, 6:75-87.

Lee TD, Genovese ED (1989). Distribution of practice in motor skill acquisition: different effects for discrete and continuous tasks. Research Quarterly for Exercise & Sport, 60(1):59-65.

Liepert J, Bauder H, Miltner WHR, Taub E, Weiller C (2000) Treatment-Induced Cortical Reorganization After Stroke in Humans. Stroke 31:1210-1216.

Lu B (2004) Acute and long-term synaptic modulation by neurotrophins. Progress in Brain Research 146:137-150.

Luft AR, McCombe-Waller S, Whitall J, Forrester LW, Macko R, Sorkin JD, Schulz JB, Goldberg AP, Hanley DF (2004) Repetitive bilateral arm training and motor cortex activation in chronic stroke - A randomized controlled trial. Jama-Journal of the American Medical Association 292:1853-1861.

Magill RA, Hall KG (1990) A review of the contextual interference effect in motor skill acquisition. Human Movement Science 9:241-289.

Mathiowetz V, Volland G, Kashman N, Weber K (1985) Adult norms for the Box and Block Test of manual dexterity. Am J Occup Ther 39:386-391.

Miltner WH, Bauder H, Sommer M, Dettmers C, Taub E (1999) Effects of constraint-induced movement therapy on patients with chronic motor deficits after stroke: a replication. Stroke 30:586-592.

Muellbacher W, Richards C, Ziemann U, Wittenberg G, Weltz D, Boroojerdi B, Cohen L, Hallett M (2002) Improving Hand Function in Chronic Stroke. Arch Neurol 59:1278-1282.

Murase N, Duque J, Mazzocchio R, Cohen LG (2004) Influence of interhemispheric interactions on motor function in chronic stroke. Annals of Neurology 55:400-409.

Nagpal A, (2005) Finger tracking training versus movement training during brain reorganization and functional improvements in home-based stroke rehabilitation, Presented at American Physical Therapy Association III STEP Conference: Linking Movement Science and Intervention, July 2005, Salt Lake City.

NIH (2004) Wearable Robots Help with Stroke Rehabilitation, Retrieved June 28, 2005, from http://www.nibib.nih.gov/eAdvances/121504.htm.

Nudo R, Milliken G (1996) Reorganization of movement representations in primary motor cortex following focal ischemic infarcts in adult squirrel monkeys. J Neurophys 75:5:2144-2149.

Nudo R, Friel K (1999) Cortical plasticity after stroke: Implications for rehabilitation. Rev Neurol (Paris) 155:713-717.

Nudo R, Milliken G, Jenkins W, Merzenich M (1996a) Use-dependent alterations of movement representations in primary motor cortex of adult squirrel monkeys. J Neurosci 16(2):785-807.

Nudo R, Wise B, SiFuentes F, Milliken G (1996b) Neural substrates for the effects of rehabilitative training on motor recovery after ischemic infarct. Science 272:1791-1794.

Oldfield R (1971) The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9(1):97-113.


Page SJ, Sisto S, Levine P, McGrath RE (2004) Efficacy of modified constraint-induced movement therapy in chronic stroke: a single-blinded randomized controlled trial. Arch Phys Med Rehabil 85:14-18.

Pascual-Leone A, Grafman J, Hallett M (1994) Modulation of cortical motor output maps during development of implicit and explicit knowledge. Science 263:1287-1289.

Pascual-Leone A, Nguyen KT, Cohen AD, Brasil-Neto J, Cammarota A, Hallett M (1995) Modulation of muscle responses evoked by transcranial magnetic stimulation during the acquisition of new fine motor skills. J Neurophysiol 74:1037-1045.

Piron L, Tonin P, Atzori AM, Zanotti E, Massaro C, Trivello E, Dam M (2002) Virtual environment system for motor tele-rehabilitation. Studies in Health Technology & Informatics 85:355-361.

Ploughman M, Corbett D (2004) Can forced-use therapy be clinically applied after stroke? An exploratory randomized controlled trial.1417-1423.

Reinkensmeyer DJ, Pang CT, Nessler JA, Painter CC (2002) Web-based telerehabilitation for the upper extremity after stroke. IEEE Transactions on Neural Systems & Rehabilitation Engineering 10:102-108.

Schmidt RA, Lee T (2005), Motor Control and Learning-A Behaviorial Emphasis, 4th ed., Human Kinetics,

Schmidt RA (1991) Frequent augmented feedback can degrade learning: Evidence and interpretations. In: Tutorials in motor neuroscience (Requin J, Stelmach G, eds), pp 59-75. Dordrecht, The Netherlands: Kluwer Academic Publishers.

Seitz RJ, Schlaug G, Kleinschmidt A, Knorr U, Nebeling B, Wirrwar A, Steinmetz H, Benecke R, H.J. F (1994) Remote depressions of cerebral metabolism in hemiparetic stroke: Topography and relation to motor and somatosensory functions. Human Brain Mapping 1:81-100.

Shea JB, Zimney ST (1983) Context effects in memory and learning movement information. In R.A. Magill (Ed.). Amsterdam: North-Holland.

Shields RK, Leo KC, Messaros AJ, Somers VK (1999) Effects of repetitive handgrip training on endurance, specificity, and cross-education. Physical Therapy 79:467-475.

Stern EB (1992) Stability of the Jebsen-Taylor Hand Function Test across three test sessions. American Journal of Occupational Therapy 46:647-649.

Taub E (1980) Somatosensory deafferentation research with monkeys: Implications for rehabilitation medicine. In: Behavioral psychology in rehabilitation medicine: Clinical applications. (Ince L, ed), pp 371-401. New York: Williams & Wilkins.

Taub E, Miller NE, Novack TA, Cook EWI, Fleming WC, Nepomuceno CS (1993) Technique to improve chronic motor deficit after stroke. Arch Phys Med Rehabil 74:347-354.

Taub, E, Uswatte, G., & Pidikiti, R. (1999). Constraint-Induced Movement Therapy: A new family of techniques with broad application to physical rehabilitation—a clinical review. J. Rehab. Res. Devel., 36, 237-251.

Taub, E, Uswatte, G. (2000). Constraint-Induced Movement Therapy and massed practice. Stroke, 31:983-c.

Tilley A (2001), The Measure of Man and Woman: Human Factors in Design, Wiley. Von Monakow C (1969) Diaschisis (1914 article translated by G. Harris). In: In: Brain and

Behavior I: Mood States and Mind. (Pribram KHe, ed), pp 27-36. Baltimore: Penguin. Weinstein S (2004), Guidelines for the Design of Home Telemedicine Systems, MS Thesis,

Department of Mechanical Engineering, University of Minnesota. Widen Holmqvist L, von Koch L, Kostulas V, Holm M, Widsell G, Tegler H, Johansson K,

Almazan J, de Pedro-Cuesta J (1998) A randomized controlled trial of rehabilitation at home after stroke in southwest Stockholm.[see comment]. Stroke 29:591-597.

Will B, Galani R, Kelche C, Rosenzweig MR (2004) Recovery from brain injury in animals: relative efficacy of environmental enrichment, physical exercise or formal training (1990-2002). Progress in Neurobiology 72:167-182.

Winters JM (2004) A telehomecare model for optimizing rehabilitation outcomes. Telemedicine Journal & E-Health 10:200-212.


Witte OW, Stoll G (1997) Delayed and remote effects of focal cortical infarctions: secondary damage and reactive plasticity. Advances in Neurology 73:207-227.

Wulf G, Shea CH (2002) Principles derived from the study of simple skills do not generalize to complex skill learning. Psychonomic Bulletin & Review 9:185-211.

Young JB, Forster A (1992) The Bradford community stroke trial: results at six months. BMJ 304:1085-1089.


Figure 1: TrackTrain home tracking system.

Figure 2: Hand sensor.


Figure 3: Interface box.

Figure 4: Calibration (left) and pre-trial (right) screens.

Figure 5: Tracking (left) and post-trial feedback (right) screens.

MAX233 LEVEL SHIFT

LEFT SENSOR

MCP6S26 MUX

MCP41100 OFFSET

2-POLE20 HZ

FILTER RIGHT

SENSOR

PWR5V VREG

9V WALL

SUPPLY

PIC16F873 CONTROLLER

BUTTO

HOST PC

COMMPORT


Table 1: Performance scores T = target trajectory, X = response trajectory, N = number of data points in record Metric and Name Computation Algorithm Interpretation E = Total Error = RMS Error 2/1

2)(1⎟⎠⎞

⎜⎝⎛ −∑ ii TX

N

RMS error between response and target.

CE = Constant Error = Average Error ∑ − )(1

ii TXN

Average error between response and target. Reveals bias above or below target.

VE = Variable Error = Standard Deviation

2/12)(1⎟⎠⎞

⎜⎝⎛ −−∑ CETX

N ii Standard deviation of the response about the bias. Measure the spread of the response.

CE/VE VECE / >1 for response far from target but consistent, <1 for response close to target but not consistent.

P = Pattern 2/12)(1⎟⎠⎞

⎜⎝⎛ −∑ TT

N i Standard deviation of target.

AI = Accuracy Index PEP /)( − RMS error normalized to the target range. AI=1 for perfect tracking.

Extension Peaks (Mean of response extension peaks)/(Mean of target extension peaks)

< 1 means response did not reach target peaks.

Flexion Peaks (Mean of response flexion peaks)/(Mean of target flexion peaks)

Same as above.

Lag Peak of cross-correlation between target and response

> 0 means response lags target.

Fade (First half RMS)/(Last half RMS) > 1 means response getting smaller as trial proceeds.

Frequency (Zero crossings of response)/(Zero crossings of target)

Large score means subject has tremor.


Table 2: Tracking block conditions Parameter Units Conditions Wave Shape n/a triangle, left-saw, right-saw, square, pseudo-random Frequency Hz 0.2, 0.4, 0.8 Duration Sec 5, 10, 15, 20 Range % of ROM 0-50, 30-70, 50-100, 0-125 Joint n/a wrist, finger Side n/a paretic, nonparetic Posture n/a pronated, supinated, mid Visual Feedback n/a on, off Table 3: Usability survey results

Questions N Strongly Disagree Disagree No

Opinion Agree Strongly Agree

1. I understood how to use the system 19 9 10 2. The computer was easy to use 19 9 10 3. I was able to put on and take off the hand sensors by myself without problems

19 1 4 10 4

4. The physical system setup in my home interfered with my daily routine

19 9 6 4

5. The scores and comments I received after each trial helped me

19 3 1 14 1

6. It was easy for me to communicate with the therapist during a tele-session

17 1 10 6

Documents

Design and Usability of a Home Telerehabilitation System ...wkdurfee/publications/Track-JBiomechEng.pdfupload tracking data to a central file server. The system was designed for use