Prototyping a Low-Cost Flexible Sensor Glove For Diagnostics and

Rehabilitation

Shival Indermun

1,2 a

and Taahirah Mangera

2 b

Mechanical and Mechatronic Department, Stellenbosch University, Western Cape, South Africa

School of Mechanical, Aeronautical and Industrial Engineering, University of Witwatersrand, Johannesburg, South Africa

Keywords:

Low-Cost, Diagnostics, Hand Rehabilitation, Flex Sensors, Prototype, Hand Impairment, Developing

Countries, Design, Evaluation.

Abstract:

Individuals in developing regions who require hand therapy for rehabilitation face difﬁculties accessing local

clinics. The objective of the current study was to create a cost-effective device capable of assessing ﬁnger

range of motion (ROM) for diagnostic and potential rehabilitation purposes in these disadvantaged areas. The

design employs ﬂexible sensors and a soft glove that records the motion of key ﬁnger joints during a variety

of daily activities performed by ten healthy participants. The results demonstrated the glove’s effectiveness

in measuring dynamic ROM for both hands of all participants. This promising outcome suggests that the

ﬂexible sensor holds great potential as a tool for hand rehabilitation and diagnosing hand impairment, offering

a valuable solution to address accessibility issues in developing countries.

1 INTRODUCTION

Low-cost prosthetic devices are well-researched,

while low-cost rehabilitative orthotic devices are less

well-addressed. Speciﬁcally, low-cost rehabilitation

of hand impairments using orthotic devices is not

commonly addressed. Many diseases and disorders

can lead to hand impairments that sees patients losing

full use of their upper limbs. One of the more preva-

lent disorders is cerebrovascular accidents, commonly

referred to as Stroke. Stroke is steadily increasing in

developing countries (Yan et al., 2016) and is one of

South Africa’s leading causes of disability (Maredza

and Chola, 2016). Stroke costs are at approximately

2-3% of the total health services expenditure in South

Africa (Maredza and Chola, 2016).

Although stroke is prominent in developing coun-

tries like South Africa, it is still one out of many that

affect the hand. Moodley (Moodley, 2018) highlights

the signiﬁcance of radial nerve palsy in South Africa.

The radial nerve accounts for the extension of the

wrist and ﬁngers, thus damage to this nerve results

in weakness and reduced mobility of the hand. The

causes of radial nerve palsy can stem from physical

injuries to infections, with the most common cause

being related to overuse of the arm. This injury is

more commonly seen in labourers where the radial

https://orcid.org/0000-0002-5569-5036

https://orcid.org/0000-0002-8113-8030

nerve could be compromised due to the extent of their

work.

Half of the Stroke patients in South Africa live in

rural areas where the nearest clinic equipped for ther-

apy is not easily accessed due to distance(Maredza

et al., 2015). Statistics South Africa (StatsSA, 2015)

found that lower-income households spend a higher

proportion of their income on public transport than

other expenses. Transport to clinics where rehabili-

tation can be accessed increases the expense of con-

tinual rehabilitation. In addition, due to the large

turnovers observed by therapists (De Klerk et al.,

2016), some patients may ﬁnd themselves waiting

for long periods of time which may result in them

needing to return the next day to receive treatment.

De Klerk et al. (2016) studied additional issues that

South Africans face in terms of occupational hand

therapy. One of the more pertinent issues is the high

rates of cases and referrals. Each therapist is restricted

to a set time for each patient. Therefore, with the

rates of cases increasing, therapists are forced to min-

imize the time allocated for consultation and treat-

ment. Limited academic resources and opportunities

for hand occupational therapy in South Africa are a

further barrier. Arguably, other developing countries

may face similar difﬁculties. Methods of treatment

that are accessible to all patients in South Africa and

the developing world are thus required.

Given the number of medical diseases that can

hinder or injure the hand, rehabilitation that can stim-

Indermun, S. and Mangera, T.

Prototyping a Low-Cost Flexible Sensor Glove for Diagnostics and Rehabilitation.

DOI: 10.5220/0012314800003657

In Proceedings of the 17th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2024) - Volume 1, pages 103-110

ISBN: 978-989-758-688-0; ISSN: 2184-4305

103

ulate recovery is imperative. However, before at-

tempting to produce methods that can provide effec-

tive therapy, consideration needs to be given to assess-

ing the level of hand impairment. Methods of mea-

suring the range of motion of the hand may result in a

means of quantifying impairment, monitoring patient

recovery and possibly diagnosing both neurological

and hand-speciﬁc disorders. Hart and Tepper (2001)

determined, through a questionnaire completed by pa-

tients with hand impairments, that patients who had

undergone rehabilitative therapy perceived improve-

ments in their functional abilities and health. By us-

ing a method of quantifying the range of motion of

the hand, the data recorded can possibly validate any

improvements observed in therapy.

Measuring the ROM of the hand has predomi-

nantly been achieved by using a goniometer. These

methods have been used in several studies (Bain et al.,

2015; Hayashi et al., 2014; Hume et al., 1990) to de-

termine the static ROM of patients performing activi-

ties of daily living (ADL). The accuracy and reliabil-

ity of the device are dependent on the patient’s ability

to hold a gesture, with minimal movements. In ad-

dition to this dependency, repeatedly measuring each

joint of each ﬁnger during activities or gestures, is te-

dious and time-consuming. This can affect the pa-

tient’s performance. Thus, a more efﬁcient method of

measuring the ROM that addressed these areas was

developed.

2 MATERIALS AND METHODS

The current study was divided into three segments:

(1) ﬂexible sensor analysis, (2) development of the

prototype, and (3) candidate testing.

2.1 Flexible Sensor Analysis

Before developing a prototype of the device, it was

necessary to analyse the ﬂexible sensor. The analysis

entailed the conﬁguration of the ﬂexible sensors and

multiple tests conducted to verify the application of

the sensors. Apart from the sensor signal drift and

the limitations, tests aimed to replicate the anatomical

variations of the ﬁnger joints of a human hand, were

included.

Flex sensors (Sparkfun Electronics, Colorado,

USA), were chosen due to their availability and cost.

Alternative sensors such as potentiometers, can be

used as they can be conﬁgured in a similar manner.

However, the difﬁculty arises when aligning the rota-

tion of the sensor in conjunction to the rotation of a

ﬁnger joint. Furthermore, designing and implement-

ing this system brings further complications when

considering the different hand sizes of patients.

The sensors are based on polymer ink and conduc-

tive particles. As the sensor is bent or ﬂexed, the con-

ductive particles move further away, increasing the

path distance the applied current must travel through,

thereby increasing the resistance. By recording the

resistance at different angles of ﬂexibility or bending,

a correlation can be made against the range of motion.

Oess et al. (2012) had sampled multiple ﬂexible

sensors with respect to signal drift, comparing differ-

ences based on both type and sensor length. The re-

sults displayed a relation between signal drift and sen-

sor length, with an increased length leading towards

a decreased signal drift. In addition, the minimum

signal deviations were observed from the sensors that

had gone through a polyester over-lamination process.

This suggests that a cover medium may result in lower

variations of the signal. The current study is heav-

ily inﬂuenced over the availability of resources and

therefore conﬁned to the use of locally sourced, sin-

gle branded and sized ﬂexible sensors.

Apart from the comments made by Oess et al.

(2012), the sensor datasheet highlighted that the base

of the sensor should be supported, and no bending

should occur near the output pin of the sensors. In the

current study a 3D printed base was used to secure the

ends of the sensors.

The ﬂexible sensors were used along the joints of

the ﬁngers. Before any prototype was developed, con-

sideration had to be given to the anatomical variation

in hand sizes and shapes. As discussed, the sensors

output resistances based on the extent at which they

are bent. The greater the bend, the higher the resis-

tance. These resistances will be used to map angles

that represent the rotation of the joint. It was thus

necessary to determine the resistance ﬂuctuations ob-

served from these sensors.The basic conﬁgured cir-

cuit consisted of an Arduino Uno, one ﬂexible sensor,

a single bread board and a 100 kΩ resistor. Only the

ﬂexible sensor was changed within the circuit when

testing for repeatability.

2.1.1 Signal Stability

The ﬂexible sensors were conﬁgured with a resis-

tor to replicate a voltage divider. This allowed the

analog pins of the selected microcontroller to read

a variable voltage. It was also necessary to deter-

mine whether the straight resistances of various ﬂex-

ible sensors were similar, had minimal signal ﬂuctu-

ations and are around 30 000 Ω, as indicated by the

manufacturer. To ensure that the readings of the sen-

sors remain undisturbed by any movements, a simple

rig was created to keep the sensor in a straight posi-

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

104

Figure 1: Flat Test Rig.

tion. This is shown in Figure 1. The straight test rig

holds the sensor in place using a 3D printed ﬂexible

sleeve. The sleeve is attached to a solid 3D printed

base using velcro.

Each ﬂexible sensor had a consistent straight line

with minor ﬂuctuations, as shown in Figure 2. This

was repeated 3 times across each sensor, resulting in

a negligible, calculated coefﬁcient of variance of less

than 1%.

Figure 2: Flat Resistances.

2.1.2 Anatomical Variations

The current investigation represented the anatomical

variations observed across patients, through a bend lo-

cation and radius test. These tests were used to deter-

mine the effect of anatomical variation on the output

of a ﬂexible sensor.

The bend location test indicated whether readings

across the length of the sensor were consistent. This

was signiﬁcant as some patients may have hands of

similar size, however the length and location of their

joints may vary. In addition, the results also deter-

mined the required positioning of the sensors in order

to produce stable outputs. Multiple ﬂexible sensors

were bent at several different locations. The locations

were determined by three offsets (20, 30 and 40 mm),

from the origin O , as seen in Figure 3.

The bend location test was repeated three times on

several sensors, to determine whether a trend was ob-

served. However, all of the sensors produced results

of varying magnitude irrespective of the offset. Fig-

ure 4 represents the three bend locations tests at a 30

Figure 3: Radius and Bend Location Test.

mm offset for a sample ﬂexible sensor. All three tests

resulted in a variation of magnitude. A coefﬁcient of

variance of less than 2.2% was calculated.

Figure 4: Repeatable Tests for Flex Sensor A (30mm Off-

set).

For the 40 mm offset (bending occurring close to

the base of the sensor) higher variations occurred, up

to 4.6%. Thus, the sensors in the current research

were positioned away from the base, towards the cen-

tre of the sensor.

Considering the structure of the hand, physical

variations may be present along the joints of the ﬁn-

gers. Some people may have bony knuckles and joints

while others may have rounded and smoother joints.

Nevertheless, it is necessary to replicate such scenar-

ios by bending the ﬂexible sensors at various radii

(r = 5, 10 and 12 mm). The test setup was similar

to that of the bend location test, shown in Figure 3,

by keeping a zero offset and using different radi rigs.

The results showed no signiﬁcant trends, apart from

ﬂuctuations in magnitude, with no deﬁnitive correla-

tion with the radii. However, considering the variation

in the resistances across the multiple sensors a maxi-

mum coefﬁcient of variance was calculated as 3.7%.

This difference is greater than the difference evaluated

in the bend location test, suggesting that the curvature

of bending the sensors has a greater effect than the

location at which the sensors are bent.

Based on the two tests, it was evident that the sen-

sors can produce repeatable results with deviations

being limited to under 4%. However, a calibration

Prototyping a Low-Cost Flexible Sensor Glove for Diagnostics and Rehabilitation

105

phase was introduced to the study to minimize the de-

viation as well as cater for signal drift, which will be

discussed in the next section.

2.2 Development of the Prototype

2.2.1 Design

The device design involves two main aspects: the

glove attachment and the circuit design. Golf gloves

were chosen for simplicity and availability, eliminat-

ing the need for manufacturing. They also facilitated

easy size selection for both right and left-handed indi-

viduals. Various attachment methods were evaluated,

including velcro and 3D printed sleeves, which were

either glued or sewn onto the glove.

While the velcro method offered modularity for

sensor attachment, it posed issues with separation

during repeated ﬁnger bending, potentially affecting

sensor readings. On the other hand, the 3D printed

sleeves effectively secured the sensors regardless of

attachment method.

Initially, ﬂexible sleeves were placed only on the

proximal interphalangeal (PIP) joints, providing rea-

sonable comfort. However, when extended to both

PIP and metacarpophalangeal (MCP) joints, it re-

sulted in increased resistance to motion, potentially

causing discomfort or harm to hand-impaired users.

The PIP joint is situated in the middle section of the

ﬁnger, while the MCP joint is located at the knuckle.

Due to the discomfort the ﬂexible sleeve attachment

was deemed impractical. To address this, medical

tape was used to secure the ﬂexible sensors, reducing

joint resistance and minimizing manufacturing costs

(see Figure 5).

Figure 5: Medical Tape Prototype.

Figure 6 represents the ﬁnal circuit that was de-

signed for the prototype. The Arduino Uno was

replaced with a different microcontroller (Micro-

robotics, Johannesburg, SA) that is similar to the Ar-

duino MEGA but is more compact. This is due to the

number of analogue pins required to connect all 10

sensors. However, the circuit in Figure 6 acts as an

example of how a single sensor is connected, but still

incorporates the main components of the circuit used

within the prototype.

Figure 6: Final Circuit.

For the testing procedure, the user wears the glove

and lays the hand on a ﬂat surface in a ﬂat position.

The ﬁrst calibration phase begins, with Push Button

1 pressed. The sensor readings are recorded and are

used as the 0° reference.

After 30s has elapsed the user changes the gesture

into a ﬁsted position for the second calibration. The

sensor readings recorded during this are used as a ref-

erence for 90°. The user is expected to hold the ﬁsted

position after the second calibration phase. This step

is used to record a set of data and compare these val-

ues with the calibration results. This procedure aims

to quantify the physical slack in the glove, signal drift

and possible anatomical variations by calculating the

difference between the two sets of data for each joint.

Thus, the difference will be used as a unique correc-

tion factor for each candidate.

LEDs acted as visual cues for each phase. All

LEDs remained off after calibration 2, signalling the

commencement of activity testing in which the user

performed the directed tasks, while the glove records

the ROM of the hand.

2.2.2 Validation

The validation of the glove consisted of testing each

joint at 30°, 45° and 60°. This was done by using a

ﬁnger goniometer. The results of the validation are

shown in Table 1. Each set of tests were repeated

three times across 3 different sensors, similar to that

of Section 2.1.

The RMSE (Root Mean Square Error) is a mea-

sure of the average of the differences between values

predicted and the values observed. The average error

across the left and right hands is 4.8 and 5.2, respec-

tively. These errors do not consider the correction fac-

tor and thus the signal drift and physical slack of the

glove.

The average deviations shown in Table 1 were cal-

culated including the correction factor as discussed in

the previous section. In comparison with the RMSE

error, it was evident that the signal drift and slack of

the glove produced additional error and thus the cor-

rection factor had to be applied. The results indicate

an accuracy of ± 5° which was congruent with re-

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

106

Table 1: Validation Results.

Average Drift Right(°) Left(°)

RMSE 5.2 4.8

Deviations 3.9 3.6

search conducted by Oess et al. (2012). The accuracy

of the glove is within an acceptable range for its ap-

plication, namely the measurement of large changes

in the ROM of the patients.

2.3 Candidate Testing

2.3.1 Study Participants

A total of 10 healthy participants were recruited. Par-

ticipation was completely voluntary with the recorded

data to remain disclosed through the use of num-

ber proﬁling. All participants were right-handed and

male, however this was not due to a selection crite-

rion, as the only requirement was the ﬁt of the glove.

Both hands were tested across the same set of activ-

ities. This study was authorized and approved from

the Human Research Ethics Committee (Medical) at

the University of Witwatersrand. Both the principle

investigator and each participant signed a declaration

of consent before each test session.

2.3.2 Performing ADLs

To determine the dynamic range of motion of an in-

dividual participant as an active member of society,

testing must include activities that encompass the ex-

pected daily routine of said individual. Gracia et al.

(2017) investigated the suitability between the active

range of motion and the functional range of motion of

the dominant hand, during various activities of daily

living. The selection of the activities was based on the

International Classiﬁcation of Functioning, Disability

and Health (ICF). The ICF is a basis for measuring

the level of health and disability for an individual or

a population. The ICF categories selected by Gracia

et al. (2017) were as follows: communication, mobil-

ity, self-care, and domestic life. By adhering to the

ICF chapters in the activity selection process, the cur-

rent study will be consistent with previous research.

Each candidate was required to perform a set of ac-

tivities derived from ICF categories using each hand.

The categories and corresponding activities selected

can be seen in Table 2.

The content of the study was explained to each po-

tential participant and an assessment of the ﬁt of the

glove was made to determine whether a participant

was suitable. Participants then completed a question-

naire to determine hand dominance and whether the

participant had any relevant medical history.

Table 2: Activities measured in each ICF category.

ICF Category Action

Self-care Brushing teeth

Buttoning a shirt

Tying a shoelace

Pouring liquid

Drinking water

Eating with a spoon

Cutting with a knife

Eating with a fork

Mobility Holding a ball

Placing a ball in a cup

Flip a card

Open a lock

Technology and

Communication

Turning pages of a book

Typing numbers on a phone

Typing

Writing

Domestic Spray a white board

Wipe a white board

The principal investigator ensured the ﬂexible sen-

sors were correctly positioned and secured before be-

ginning calibration. During calibration, the partici-

pant was required to keep their hand in a ﬂat posi-

tion on top of a desk. This represents the 0

◦

position,

which was recorded by the device. Thereafter, the

participant gestured a speciﬁc ﬁst, while ensuring that

all joints were as close to 90

◦

as possible. The prin-

cipal investigator closely investigated and corrected

the required hand positions. Once more, the data was

recorded. These two streams of data were averaged

and mapped to 0

◦

and 90

◦

, respectively.

The participant was directed to perform a certain

activity. Once observed, the candidate repeated the

activity while the device recorded the data. Follow-

ing data recording on both hands, an evaluation was

completed by the participant, indicating the level of

comfort while using the glove and any overall com-

ments that they may have had about the study.

3 COST

Components were sourced locally. The ﬂexible sen-

sors were the highest-cost component (61% of the to-

tal cost), followed by the golf gloves and the micro-

controller. This is dependent on the number of sen-

sors used which was 10 in this study. Since this is

a critical component, cost reduction must be directed

elsewhere.

Nylon or nitrile gloves are non-stretch and will

thus simplify attaching the sensors to the gloves.

These therefore present a lower-cost alternative.

The resulting material cost of the prototype was

approximated to $121. Comparing the costs of the

Prototyping a Low-Cost Flexible Sensor Glove for Diagnostics and Rehabilitation

107

goniometer and the glove is not straightforward solely

based on their price. The goniometer, while more af-

fordable, involves a labor-intensive process for ther-

apists and may be particularly challenging for hand-

impaired patients. This could potentially lead to ex-

tended therapy sessions and higher overall costs in

terms of time and resources.

In contrast, the glove offers a more efﬁcient

and technologically advanced approach. It not only

streamlines the measurement process, making it less

burdensome for both therapists and patients, but it

also introduces mechatronics into the rehabilitation

process, aligning with the broader goal of integrating

technology into local and rural healthcare practices.

4 RESULTS

The dynamic tests were successfully conducted on 10

participants. Both the dominant and non-dominant

hands were measured across a set of 20 activities.

In Figure 7 (a) and (b), the ROM measured is de-

picted for two activities: (a) placing the ball into the

cup (Candidate 1 - left hand) and (b) pressing a spray

bottle (Candidate 10 - right hand). A wooden hand

model visually depicts the ROM by taking a few ref-

erence points from the graphs to position the hand

model.

4.1 Dynamic ROM

Figure 7(a) represents a collected motion of all the

ﬁnger joints while Figure 7(b) describes an action that

can be explained by the movements of a single joint

(PIP joint of the index ﬁnger).

4.2 Dominant vs. Non-Dominant Hand

and Joint Variation

An approximate of the amount of change observed be-

tween the joints can determine which joint or hand has

a greater degree of controlled motion. Based on this

premise, the coefﬁcient of variance was calculated for

each hand and each joint for every participant, across

all activities. These values were then summed for

each participant to represent the total variation for

each hand. The total variation of both hands can be

seen in Figure 8.

Considering all participants as a whole, the varia-

tion amongst the joints were summed and are shown

in Figure 9.

Figure 7: (a) Candidate 1: Placing Ball Into a Cup, (b) Can-

didate 10: Using a Spray Bottle.

5 DISCUSSION

All 10 candidates performed all of the required activi-

ties with ease. The results produced from the study

were successful in recording the dynamic range of

motion. From Figure 7(a), all 9 joints began to in-

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

108

Figure 8: Total Hand Variation of all Candidates.

Figure 9: Total Variation across each Joint.

crease in ROM from time t = 1.5-2.5s. Thereafter a

slow decrease was observed until time t = 5.5 s, fol-

lowed by a rapid reduction in ROM. This behaviour

described the participants physical movements. For

example, from times t = 0 - 1.5 s (A) the candidate

was reaching for the ball. From times t = 2.5 - 5.5 s

(B) the ball was then picked up and placed into the

cup. Finally, from time t = 5.5 s (C) onwards the par-

ticipant had released the ball and their hand was back

in a neutral position.

Figure 7(b) describes an action that can be ex-

plained by the movements of a single joint. The task

was the action of pressing a spray bottle. The graph

shows two stages, times t = 1.5 -3 s (A), t = 3 - 4.5

s (B) corresponding to a dip and rise in the ROM of

the PIP joint in the index ﬁnger. These stages can be

seen as the pressing of the head on the spray bottle.

According to the ROM of all the joints, the spray bot-

tle is released from time t = 5 s onwards. The hand

model represents the changes of the PIP Joint.

Being able to correlate the results to a hand model

that resembles the requested activities can validate the

application of the glove as a device to monitor the

ROM of the hand. Previous studies have used ﬁn-

ger goniometers (Bain et al., 2015; Hume et al., 1990;

Hayashi et al., 2014) as they are reliable devices for

joint measurements. However, they will require a pa-

tient to hold the requested gestures for longer peri-

ods as the therapist needs to measure each joint of

each ﬁnger. This may be difﬁcult for hand-impaired

patients. Irrespective of the duration of the measure-

ments, the ﬁnger goniometer is targeted towards static

hand gestures while the current study is focused on

dynamic motions. Each activity was performed in

10 s with a recording every 0.5 s. By manipulating

the recording times, a smoother representation of the

hand motion can be attained. Thus, replacing the in-

stantaneous drops observed in the various plots with

a more detailed reduction in ROM.

It was expected that most participants would have

a greater variation in their right hands due to their

right-hand dominance. This can be seen in Figure

8. However, this was not the case for Candidate 7

and 10. This can be due to over-gripping gesture, ob-

served during the testing phase, of the non-dominant

hand when attempting the various activities as a result

of lack of control.

The IP joint of the thumb experiences the most

variation in both the left and right hands of all par-

ticipants (Figure 9). This is expected, as the thumb

is what separates humans from most animals because

it allows for complicated hand gestures. The thumb

promotes functionality in the forms of gripping and

pinching, which can correlate to the majority of the

tasks within the current study. Excluding the thumb

and summing the variation across each ﬁnger showed

that the maximum variation occurred in the little, in-

dex, middle and ring ﬁnger, in order of highest to low-

est. This trend was present in both the right- and

left-hand results. The little ﬁnger does not domi-

nate motion with respect to functionality however it

does achieve a full ROM due to its closed positioning

throughout the fundamental movements. For exam-

ple, when gripping or pinching a pencil from a neutral

position, the little ﬁnger would move from an opened

to a completely ﬁsted position. This is due to the

thumb, index and middle ﬁnger dominating the ac-

tion. The index and middle ﬁnger support the thumb

in functional movements of the hand and thus exhibit

the next highest variation.

For the evaluation of the glove, the average rat-

ing (between 1-5, 5 being the most discomfort) was

2.02. Therefore, the prototype was mostly comfort-

able, throughout the procedure. However, this rating

is with respect to healthy candidates. Any level of

discomfort indicated by healthy participants could be

magniﬁed in the case of a patient suffering with hand

impairment.

The reasoning of the slight discomfort had to do

with the sizing of the glove. Expanding the study will

require gloves of varying sizes. While the current de-

vice acts as a prototype, newer designs must be de-

veloped to cater for variation in hand size, but most

importantly patient comfort.

Prototyping a Low-Cost Flexible Sensor Glove for Diagnostics and Rehabilitation

109

6 CONCLUSIONS

The ﬂexible sensor glove proved capable of measur-

ing the dynamic range of motion for each hand of all

10 participants. A hand model was positioned accord-

ing to the dynamic plots and resulted in replicated

hand gestures that would have occurred during the

speciﬁc activities. Being able to correlate the data to

the speciﬁc activity, conﬁrms the capability of gloves

in measuring the dynamic ROM.

Comparisons were made between the dominant

and non-dominant hand of some participants. The re-

sults showed a greater ROM in the dominant hand for

most candidates. However, a few participants expe-

rienced a higher ROM in their non-dominant hand.

Based on these results and observations made during

the testing procedure, it was evident that some partic-

ipants would exaggerate their grip onto objects due to

lack of control with their non-dominant hand.

The variation of the motion between joints was

calculated using the coefﬁcient of variance. The vari-

ation was more prominent amongst the right hand and

therefore could suggest dominance. The IP joint of

the thumb had a maximum variation of ROM for both

the left and right hand. The thumb is a signiﬁcant joint

with respect to hand functionality and thus would ob-

serve a higher ROM throughout the activities.

According to the participants, the glove was mod-

erately comfortable, however new designs must be

implemented for the application on hand-impaired pa-

tients to achieve a maximum level of comfort and

safety. Apart from the design aspect, future work

should include a larger test scale, while introducing

static gestures for comparisons with previous work.

The prospective applications of the glove include

diagnostics, patient monitoring and rehabilitation. In

the case of diagnostics and patient monitoring the

glove can act as an aid to quantify the level of im-

pairment. Rehabilitation methods can incorporate

the glove with virtual reality systems or exoskele-

tons in the form of bilateral therapy. Speciﬁcally,

the glove holds promise in aiding individuals facing

challenges associated with stroke, radial nerve palsy,

tendinopathies and similar pathologies. However, for

the glove to act as an effective tool clinically, the study

needs to implement the above recommendations as

well as further testing.

REFERENCES

Bain, G., Polites, N., Higgs, B., and et al. (2015). The func-

tional range of motion of the ﬁnger joints. J Hand

Surg Eur Vol, 40(4):406–411.

De Klerk, S., Badenhorst, E., Buttle, A., and et al. (2016).

Occupation-based hand therapy in south africa: Chal-

lenges and opportunities. S Afr J Occup Ther,

46(3):10–15.

Gracia, V., Vergara, M., Sancho-Bru, J., and et al. (2017).

Functional range of motion of the hand joints in activ-

ities of the international classiﬁcation of functioning,

disability and health. J Hand Ther, 30:337–343.

Hart, D. L. and Tepper, S. (2001). Changes in health status

for persons with wrist or hand impairments receiving

occupational therapy or physical therapy. American

Journal of Occupational Therapy, 55(1):68–74.

Hayashi, H., Shimizu, H., Okumura, S., and et al. (2014).

Necessary metacarpophalangeal joints range of mo-

tion to maintain hand function. Hong Kong J Occup

Th, 24(2):51–55.

Hume, M., Gellman, H., McKellop, H., and et al. (1990).

Functional range of motion of the joints of the hand. J

Hand Surg Am, 15(2):240–243.

Maredza, M., Bertram, M., and Tollman, S. (2015). Disease

burden of stroke in rural south africa: An estimate of

incidence, mortality and disability adjusted life years.

BMC Neurology, 15(1).

Maredza, M. and Chola, L. (2016). Economic burden of

stroke in a rural south african setting. eNeurologi-

calSci, 3:26–32.

Moodley, D. (2018). Radial nerve injuries. Interview - Oc-

cupational Therapist at Charlotte Maxeke Johannes-

burg Academic Hospital, SA.

Oess, N., Wanek, J., and Curt, A. (2012). Design and evalu-

ation of a low-cost instrumented glove for hand func-

tion assessment. J NeuroEng Rehabil, 9(1):1–11.

StatsSA (2015). Measuring household expenditure on pub-

lic transport. http://www.statssa.gov.za/?p=5943 - Ac-

cessed 28 January 2019.

Yan, L., Li, C., Chen, J., and et al. (2016). Preven-

tion, management, and rehabilitation of stroke in

low and middle-income countries. eNeurologicalSci,

2(8):21–30.

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

110