Personalising Exergames for the Physical Rehabilitation of Children

Affected by Spine Pain

Cristian G

omez-Portes

, Carmen Lacave

, Ana I. Molina

, David Vallejo

and Santiago S

anchez-Sobrino

School of Computer Science, University of Castilla-La Mancha, Paseo de la Universidad 4, 13071 Ciudad Real, Spain

Keywords:

Rehabilitation, Exergames, Spine Pain, e-Health.

Abstract:

Injuries or illnesses related to the lumbar spine need great clinical care as they are one of the most prevalent

medical conditions worldwide. The use of exergames has been widespread in recent years and they have been

put forward as a possible solution for motivating patients to perform rehabilitation exercises. However, both

customizing and creating them is still a task that requires considerable investment both in time and effort. In

this project we present a language with which we have designed a system based on the physical rehabilitation

of patients suffering from bone-marrow injuries, which enables customization and generation of exergames.

To assess the system, we have designed an experiment with an exergame based on the physical rehabilitation

of the lumbar spine. The purpose of this was to assess its understanding and suitability, whose result reveals

that the tool is fun, interesting and easy to use. It is hoped that this approach can be used to considerably

reduce the complexity of creating new exergames, as well as supporting the physical rehabilitation process of

patients with lower back pain.

1 INTRODUCTION

In a systematic analysis (Murray et al., 2012) of the

research, called, The Global Burden of Disease 2010,

it was estimated that lower back pain is one of the

10 most frequent injuries and illnesses worldwide. It

has been calculated that every year between 6.3% and

15.4% of people suffer from lower back pain, while

any type of this complaint varies between 1.5% and

36% per year (Hoy et al., 2010). Moreover, in mon-

etary terms, treating lower back pain complaints is

an enormous burden on industries and governments,

which is even more acute for patients and families

dealing with them (Duthey, 2013).

Unfortunately, this type of condition is rather

frequent in children and teenagers between 10 and

16 (Jones et al., 2004), who need to perform rehabil-

itation exercises, which have been recommended by

a physiotherapist, at home, and that are essential for

reducing the area of pain.

Today, there are some relatively new treat-

https://orcid.org/0000-0002-9603-9481

https://orcid.org/0000-0003-2770-8482

https://orcid.org/0000-0002-3449-2539

https://orcid.org/0000-0002-6001-7192

https://orcid.org/0000-0001-6620-1719

ments for physical rehabilitation, the main aim of

which are to improve the quality of life of the pa-

tients (O’Sullivan et al., 2019). Among these are

tele-medicine-based solutions, aimed at making these

types of tasks accessible and providing patients with

a greater degree of independence (Palacios-Navarro

et al., 2015; Lai et al., 2015). There is even a line of

research in which gamiﬁcation-based techniques and

serious games are used, in order to create technolog-

ical solutions which have motivational value to en-

courage patients to perform more rehabilitation tasks

or to enjoy them whilst reducing their pain (Deterding

et al., 2011; McCallum, 2012).

However, these approaches are not enough in

themselves to tackle two challenges: i) motivating pa-

tients to continuously carry out the programme of ex-

ercises assigned to them and ii) assessing whether the

exercises have been performed correctly. Generally, a

child or teenager will perform the exercises at home,

but without the necessary motivation, the routine re-

quired to do them will fade, and thus the desired thera-

peutic effect will be lost. Moreover, from the point of

view of the physiotherapist, it is advisable to, ﬁrstly,

have an application or game which encourages the pa-

tient to perform the exercises so that he or she can be

automatically guided, and to ensure the exercises are

carried out correctly. In other words, to provide cus-

Gómez-Portes, C., Lacave, C., Molina, A., Vallejo, D. and Sánchez-Sobrino, S.

Personalising Exergames for the Physical Rehabilitation of Children Affected by Spine Pain.

DOI: 10.5220/0009574005330543

In Proceedings of the 22nd International Conference on Enterprise Information Systems (ICEIS 2020) - Volume 2, pages 533-543

ISBN: 978-989-758-423-7

533

tomized activities according to the patients and their

type of illnesses so as to obtain a therapy that is both

safe and effective (Pirovano et al., 2016).

In this research, the design, developing and val-

idating a system focused on rehabilitating children

and teenagers suffering from some sort of lower back

pain, is set out, making use of development kits

with advanced artiﬁcial intelligence sensors, such as

Azure Kinect DK

device, so as to accurately track

skeletons. In this way, the patient naturally inter-

acts with the system, simulating rehabilitation exer-

cises by means of motivating games. The rehabili-

tation tasks are customized with a language we have

deﬁned as Personalised Exergames Language (PEL)

which, on the basis of the GL Transmission Format

speciﬁcation (glTF) (Robinet et al., 2014), keep both

information about the exergame mechanics, gamiﬁca-

tion elements and metrics for measuring how the pa-

tients are progressing. This information is deﬁned by

therapists based on the patient’s condition, which is

converted into PEL sentences for automatically gen-

erating exergames. Therefore, a co-creative approach

is established between therapists, patients and devel-

opers.

By way of validation, the system has been as-

sessed by 23 potential users, being the main objective

to obtain feedback about its understanding and suit-

ability.

The remainder of the article is structured as fol-

lows: The section 2 positions our research within the

context of other works in the ﬁeld. Then, the sec-

tion 3 sets out the system architecture. Subsequently,

the section 4 describes the experiment carried out and

the results obtained. Finally, in the section 5, the con-

clusions drawn are described as well as future lines of

research.

2 RELATED WORK

At present, there are tools for rehabilitation whose ex-

ercises are automatically analysed by machines which

use precise skeleton tracking or computer vision tech-

niques. The latter have traditionally been based on

using the Kinect device (Webster and Celik, 2014;

Da Gama et al., 2015), a low cost hardware solution

whose effectiveness has been seen in the ﬁeld of phys-

ical rehabilitation (Clark et al., 2012; Mobini et al.,

2014; Mousavi Hondori and Khademi, 2014). Simi-

larly, there are solutions on the market which are ori-

entated at training and ﬁtness, whose aim is to attract

speciﬁc targets, apart from providing remote rehabili-

tation from home (Deutsch et al., 2009; Esculier et al.,

2012). However, some authors shows scepticism to-

wards this approach, claiming that the lack of cus-

tomization of these games may make injuries more

likely (Sparks et al., 2011).

In light of this issue, various authors have set out a

series of considerations to ensure that exergames are

both efﬁcient and safe. In (McCallum, 2012), one of

the most outstanding proposals is to control the game

experience, given that, depending on whether the tar-

get user is a child or adult, the original concept of

the game may be misinterpreted. A similar approach

is presented in (Wiemeyer et al., 2015), which pro-

vides recommendations for the optimal design of ex-

ergames. To take an example, it identiﬁes postures

which ensure patient safety when in rehabilitation,

or adapts the game design to the patient characteris-

tics. A slightly different approach, which is presented

in (Pirovano et al., 2016), provides a methodology

split into four phases for creating and designing ef-

ﬁcient and safe therapeutic exergames.

The use of exergames and gamiﬁcation techniques

has also been studied in the literature to assess their

effectiveness in the rehabilitation process. A system-

atic review, shown in (Matallaoui et al., 2017), an-

alyzes different types of systems and their context,

the game elements used and the results yielded from

them, providing positive outcomes both in patient

conduct and their well-being. In a similar vein, the

research presented in (Gonz

alez et al., 2018) explores

studies on games and gamiﬁcation applied to physical

exercise, which ﬁnds positive effects in a reduction

in body weight and in encouraging physical exercise.

Similarly, the paper (Katajapuu et al., 2017) shows a

set of exergames evaluated by a group of thirty partic-

ipants, which concludes that video games are useful

in the physical training of elderly people.

There is also a state of the art angle in which

technological solutions orientated at facilitating or au-

tomating the generation of exergames is envisaged.

For example, the TANGO:H (Tangible Goals:Health)

platform (Gonz

alez et al., 2013), has a range of func-

tions and contains a graphics editor to help experts

create exercises adapted to patient needs. In this

case, it refers to hospitalized children. Similarly,

in (Hardy et al., 2015), the StoryTec tool has been

envisaged. This was designed to support experts in

the ﬁeld, such as doctors or therapists, so that they

can adapt and customize games-based training pro-

grammes for elderly and disabled people. In a sim-

ilar vein, the VirtualGym environment is also worth

mentioning (Fernandez-Cervantes et al., 2018), which

has been set out as a cooperative framework in which

medical professionals design exercise routines which

later become games in which an avatar guides the pa-

tient during his or her rehabilitation process.

ICEIS 2020 - 22nd International Conference on Enterprise Information Systems

534

3 ARCHITECTURE

Figure 1 graphically shows the architecture designed

for the rehabilitation process for bone-marrow in-

juries. This approach mainly requires an expert in the

ﬁeld to participate, namely, a therapist, entrusted with

assigning an exercises routine to the patient. As can

be seen in the ﬁgure, the exercises routine is deﬁned

by means of data modelling in .gltf format extended

with our PEL language, which stores all information

relating to the exergame. That is, the game dynam-

ics, graphics, gamiﬁcation elements, and the different

metrics by which patient participation can be evalu-

ated. In the subsection 3.1, there is an in-depth expla-

nation on the structure of the language we have de-

ﬁned, so as to provide details in the subsection 3.2

on the translation process for this model in order to

automatically generate exergames.

Additionally, this architecture is based on a hard-

ware component for precise skeleton tracking, here

we use Azure Kinect DK

, a device developed by

Microsoft which includes a RGB (Red, Green and

Blue) camera, a depth sensor and several micro-

phones, apart from the software necessary for cap-

turing movements and recognizing voice commands.

The capture module is that which interacts with the

hardware device in order to obtain the information as-

sociated with the position and orientation of the bones

that make up the patient skeleton. This module is re-

sponsible for ﬁltering the data provided by the device,

excluding any information that is not related to the

bones in the rehabilitation exercise. For example, if

patients have to lie down to exercise their backs, and

raise the upper part of their torsos, data related to the

lower part of the hip is omitted. Apart from that, the

capture module also manages voice recognition, as it

enables the voice of the patient to be analysed in order

to perform certain tasks, namely, starting a rehabilita-

tion routine or indicating that the activity has ﬁnished.

The processing module carries out two speciﬁc

tasks: i) it evaluates the exercise performed by the

patient and ii) it monitors motivation according to

the activity. A games scenario by deﬁnition in our

system is made up of an avatar and a sequence of

nodes, called actors, which make up the path a spe-

ciﬁc part of the patient’s body must take. To evaluate

the exercise, the module analyses the set of bones that

have interacted with the sequence of actors and the

order in which this interaction has taken place. In this

way, patients are aware of whether they are perform-

ing the exercise correctly, as the module shows scores

as the activity progresses. Furthermore, it helps to

keep them motivated, whether the exercise is being

performed correctly or not, since a high score may

help maintain the rhythm, or even a low score may

help incentivize them to improve.

Finally, the visualization mode is responsible for

showing on the screen how the patient is progressing

in respect to the task performed. This module is in-

terconnected with the processing module, which pro-

vides information in numerical format (score the pa-

tient has, percentage of progress for the activity or

achievements, amongst other items) in a visually at-

tractive format which can easily be interpreted by the

patient in order to capture and maintain their motiva-

tion and attention.

3.1 PEL Structure

PEL comes from a higher level speciﬁcation, pop-

ularly known as glTF

, a relatively new, open for-

https://www.khronos.org/gltf/

Figure 1: General overview of the proposed architecture for rehabilitation of patient suffering from bone-marrow injuries.

Personalising Exergames for the Physical Rehabilitation of Children Affected by Spine Pain

535

mat, based on the JSON standard, for distributing 3D

scenes in an efﬁcient and interoperable way. It was

chosen because i) it is an open project, which means

certain parts of the product can be freely changed to

customize it; ii) it is an efﬁcient and interoperable for-

mat; and iii) the extensibility of its data modelling

by which new properties can be added which provide

new opportunities for improving and building on the

speciﬁcation.

Fundamentally, our language is made up of a se-

ries of components that potentially enable tools to be

developed, which through parsing, automatically gen-

erate games-based exercises focused on rehabilitating

patients. The following points brieﬂy summarize the

elements which deﬁne the nature of an exergame with

PEL (see Figure 2):

Figure 2: Representation of an exergame structure using

glTF extended with PEL language.

• Views. Contains three different views which the

game is split into. That is, the language envisages

a tutorial view, where an avatar, using animation,

shows the patient the activity to be carried out; a

scene in which the patient performs the activity

that has been visualized previously, with move-

ments replicated by the avatar; and ﬁnally a view

where the results are shown as a result of the ac-

tions carried out by the patient.

• Actors. They represent the elements of the game

that the avatar must interact with. These objects

essentially make up a sequence of the activity the

patient must perform, with behaviour that may be

either static or dynamic. That is, elements that

move in a 3D space through animation, or which

are just ﬁxed elements suspended over a certain

point in the 3D scene. Moreover, these elements

provide visual feedback, which in itself, indicates

that the node must be or has interacted with the

target joint.

• Gameplay. Deﬁnes the set of actions the user

must take to complete a repetition of the ex-

ergame. In other words, it speciﬁes the sequence

of the actors and the user interaction mode with

them. Correct execution of the game mechan-

ics deploys gamiﬁcation elements to capture, in-

terest and motivate the user, which results in in-

creases the score the patient achieves for each rep-

etition, the progress of the activity or on unblock-

ing achievements, for example.

3.2 Automatic Generation of Exergames

An exergame is automatically generated from a hier-

archical analysis of a ﬁle with a gltf extension, whose

structure is deﬁned by means of the JSON syntax.

This system has been implemented as an appli-

cation developed with Unity 3D

game engine, de-

ployed for Windows and Linux, both of them duly

supported by Azure Kinect DK

. This application re-

ceives the contents of a 3D scenario in the way of

a URI (Uniform Resource Identiﬁer), which is sub-

sequently analysed by means of parsing technique in

order to build the game.

The mechanism for adding gITF extensions by

means of the “extras” ﬁeld have made it possible to

create these type of scenarios. All the elements which

a 3D scenario is made up of (scene, node, camera, ma-

terial, animation, etc.) are added in order to provide

new functions for highly speciﬁc cases.

The process for analysing the “extras” ﬁeld is inte-

grated into the parsing module, which takes the form

of a syntax analyser, based on the gITF implementa-

tion, in order to effectively deserialize the elements

included in the JSON ﬁle. In this ﬁeld the attributes,

which form part of the PEL language, are included to

add speciﬁc behaviour so as to support the physical

rehabilitation process for patients.

To each of the “extras” ﬁeld, we have added an

identiﬁer as a primary property, whose purpose is to

help determine the type of behaviour which should be

assigned to a node (e.g., { "type" : "actor" }).

Therefore, once the object which implements an ex-

tensible ﬁeld is detected, our parsing module identi-

ﬁes the information to be deserialized so that it be-

haves appropriately when the exergame is running.

https://unity.com/es

https://azure.microsoft.com/es-es/services/kinect-dk/

ICEIS 2020 - 22nd International Conference on Enterprise Information Systems

536

4 VALIDATION

In order to evaluate the designed system, a quasi-

experiment with potential patients has been carried

out. The purpose of this was to validate the system

in terms of understanding and suitability. The results

of such experiment are described and discussed in this

section.

4.1 Exergame to Evaluate

The architectural design deﬁned in this paper supports

the integration of a wide range of exergames orien-

tated towards different types of injuries, ranging from

lower limbs to upper ones.

Taking advantage of this architecture, an ex-

ergame was developed to motivate potential patients

to perform exercises aimed at the physical rehabili-

tation of the lumbar spine. The exergame was cus-

tomized by a therapist, who deﬁned the therapeutic

objetive, the trajectory of the rehabilitation movement

and the metrics to measure the progress of the patient,

being this information translated into PEL sentences

to automatically generate the game.

This exercise consists in facing up from a lying

down position, raising the upper part of the torso

by about 25 cm with knees bent and feet resting on

the ground. Patients must maintain this position for

2 seconds and then return to their original position.

In essence, the exercise is repeated 5 times, where

the satisfactory realization of each one increases their

score by 200 points.

Figure 3 shows the different views which make up

the exergame designed for the physical rehabilitation

of bone-marrow injuries. In (1) there is a screenshot

of the tutorial view, where the avatar is seen carry-

ing out the movement, which the patient must repli-

cate later on. The participant view (2) reﬂects how

the patients repeats the movement shown in the pre-

vious view, increasing his or her score each time the

repetition is performed correctly. Note the spherical

blue object (actor), which is used both as a guide for

performing the exercise and as a mechanism for de-

termining the stopping conditions in a repetition. Pa-

tients can interact with the exergame as PEL is highly

ﬂexible and the language is versatile enough to de-

ﬁne these situations. In this case, once the head of the

avatar collides and remains on the sphere for 2 sec-

onds, the repetition ﬁnishes and the score increases.

Finally, the results view (3) reﬂects the indicators en-

tered in PEL language for measuring the exercise per-

formed by the patient.

4.2 Participants and Method

The experiment was carried out by two instructors and

some girls from a rhythmic gymnastics club. Twenty-

three girls were randomly selected, of whom 20, aged

between 11 and 19, agreed to participate in the expe-

rience. Of these, 8 suffered from some type of injury,

mainly related to the spine, which requires rehabil-

itation exercises at home. In order to avoid biasing

the results (McCambridge et al., 2014) and to mo-

tivate their participation (Shull et al., 2007), we ex-

plicitly stated at the beginning of the experience the

information collected would be treated conﬁdentially

and used exclusively for this research. After being in-

formed, girls and their parents gave their consent to

Figure 3: Rehabilitation system in action. (1) tutorial view. (2) Participation view. (3) Result view. (4) Features of the Azure

Kinect DK

device. (5) User using the system.

Personalising Exergames for the Physical Rehabilitation of Children Affected by Spine Pain

537

use their data.

The quasi-experiment was divided into two

phases, conducted in one session of 120 minutes:

• Phase 1 (Preparation Phase). One instructor

presented the system to all participants for ten

minutes. An example of using the system was

projected onto the wall so that the girls may un-

derstand the explanation.

• Phase 2 (Development Phase). Each girl par-

ticipating individually engaged in two activities:

First, they completed the prepared exergame,

which required about 2 minutes; then, they ﬁlled

in a questionnaire, which was provided to them

through One Drive Forms. These concerned their

perception of the activity developed, their under-

standing and how suitable they thought the tool

was.

The exergame, used as an exercise for rehabilitat-

ing the spine, was projected onto the wall so that the

girls may better visualize their avatar, as well as the

gamiﬁcation elements provided by the system.

The questionnaire consisted in 21 items, shown in

Table 1, rated on a ﬁve-point Likert scale (1: totally

disagree; 5: totally agree) grouped into ﬁve blocks

or dimensions: performance subjective ratings, cog-

nitive load and effort, system utility, usefulness of the

user interface components and TAM-based question-

naire (abbreviated by TAM). (1) The ﬁrst dimension

(performance subjective ratings), composed of four

items, allowed us to measure the subjective percep-

tion of the users regarding their performance during

the activity, assessing aspects such as their interest

during its execution, their commitment to doing it

correctly, as well as the user-friendliness of the sys-

tem. (2) The next block (cognitive load) was formed

by four items, inspired by the Cognitive Load The-

ory (CLT) (Sweller et al., 1998), which allowed us

to measure two of the types of cognitive load about

the use of a software system: the complexity im-

posed by the task to be performed (intrinsic load)

and the complexity imposed by the use of the soft-

ware and the interaction devices used during the per-

formance of the task (extraneous load). In addition,

two questions were included related to the effort the

users had to make to complete the task. (3) The

third block consisted in four questions related to cer-

tain users’ views: preference for the use of this type

of systems over face-to-face assistance in rehabilita-

tion centres, the system’s game format and ﬁnally,

whether they considered that its use might improve

their motivation and constancy in rehabilitation tasks.

(4) In the following block, ﬁve items were included

in which users had to assess the degree of usefulness

of each of the main elements of the application’s user

interface (virtual representation of the user, number

of repetitions, score, etc.). (5) Finally, users com-

pleted a questionnaire based on the Technology Ac-

ceptance Method (TAM) framework (Davis, 1993),

which included four items to measure the perceived

user-friendliness, usefulness and intended uses for the

system being evaluated.

To speed up the process of ﬁlling in the question-

naire, we used three laptops so that girls may do it in

parallel once they ﬁnished the exergame. There was

no time limit for this but, on average, each girl used

about 6 minutes to complete the questionnaire. In ad-

dition, another instructor was in charge of explaining

the meaning of some questions to the youngest girls

who requested them.

4.3 Results and Discussion

The results obtained from the data collected, illus-

trated in Table 1, show that the tool has been very

well received in all dimensions. It should be stressed

that the activity was more fun (item 1) and interest-

ing (item 2) for those girls who had some injury and,

therefore, needed rehabilitation, which seems logical.

Moreover, all participants thought they had tried to do

the activity well (item 3), they had found the system

useful and it had been user-friendly (item 18).

All participants considered that the cognitive load

of the activity was not very high (items 5, 8). Fur-

thermore, most of them tried to concentrate on the

activity (item 6). However, item 7 received a very

high score. This may be because the question was

posed in the opposite way to the others. Therefore, the

results show inconsistencies; probably because some

girls were confused when interpreting the values of

the answers.

On the other hand, the injured girls valued the

system more useful (items 9, 10) because they were

aware of what it meant for them to perform exercises

at home (without going to rehab centre). One highly

important point to consider is that the system can be

motivating (item 11), since one of the disadvantages

to rehabilitation for young people is the lack of mo-

tivation to do the exercises at home. Perhaps moti-

vation is key to understanding the greater enthusiasm

of the injured girls to using the system at home (item

20), as well as to recommending to friends (item 21).

As for, the usefulness of the interface elements

(items 13-17), they received positive appraisal with

the “score” being the best-rated one (item 15). Apart

from this, it is needed to bear in mind that gamiﬁca-

tion has a crucial role in motivating a patient, espe-

cially children.

ICEIS 2020 - 22nd International Conference on Enterprise Information Systems

538

Table 1: Descriptive statistics of the dimensions evaluated. *Mean and standard deviation are shown (in parentheses).

Dimension Item Mean Mode

Injured Not Injured Injured Not Injured

Activity Perception

1. This activity has been fun for

me.

5.00 (0.00)* 4.33 (0.65)* 5 4

2. I found this activity interesting.

4.88 (0.35)* 4.75 (0.62)* 5 5

3. I have worked to do it well.

4.88 (0.35)* 4.92 (0.29)* 5 5

4. It’s been easy for me to learn

how to use this system.

4.88 (0.35)* 4.83 (0.39)* 5 5

Cognitive Load

5. The activity required a lot of

concentration.

3.00 (0.76)* 3.00 (1.48)* 3 3

6. I’ve been very concentrated

during the activity.

3.75 (0.89)* 4.17 (1.03)* 3 3

7. I’ve had to work pretty hard to

get the activity done.

3.13 (1.36)* 2.17 (1.27)* 4 1

8. I have found difﬁcult to per-

form the rehabilitation exercise

using this system.

1.25 (0.53)* 1.50 (0.45)* 2 1

Utility

9. I’d rather use this system at

home than have to go to a rehab

centre.

4.13 (1.13)* 4.00 (1.54)* 5 5

10. This system would make me

more consistent in performing the

exercises at home.

4.75 (0.46)* 4.25 (1.06)* 5 5

11. I believe that using this sys-

tem to do rehabilitation exercises

can be motivating.

5.00 (0.00)* 4.67 (0.49)* 5 5

12. I like the application has the

format of a game.

4.50 (1.41)* 5.00 (0.00)* 5 5

Interface Elements

13. The design of the avatar is ap-

propriate.

4.88 (0.35)* 4.42 (1.24)* 5 5

14. The information about the

repetitions is useful.

4.50 (1.41)* 4.58 (1.00)* 5 5

15. I like to get score every time I

perform a good exercise.

5.00 (0.00)* 4.58 (0.79)* 5 5

16. The route composed of

spheres helps to perform the ex-

ercises.

4.88 (0.35)* 4.58 (1.16)* 5 5

17. The gym has immersed me in

a rehabilitation environment.

4.88 (0.35)* 4.75 (0.45)* 5 5

TAM

18. This system is easy to use. 4.88 (0.35)* 4.92 (0.29)* 5 5

19. Using this system could help

me in performing the rehabilita-

tion exercises.

5.00 (0.00)* 4.83 (0.39)* 5 5

20. If I could borrow this system,

I would use it at home.

5.00 (0.00)* 5.00 (0.00)* 5 5

21. I’d recommend my friends to

use this system to do the exercises

at home.

5.00 (0.00)* 4.83 (0.39)* 5 5

Personalising Exergames for the Physical Rehabilitation of Children Affected by Spine Pain

539

Table 2: Signiﬁcant correlations among items (*:p<0.05; **:p<0.01).

Item 5 Item 10 Item 11

Item 14 Item 15

Item 16 Item 17

Item 1

.457*

Item 2

.468* .509* .498*

Item 6

.584** .427*

Item 9

.520*

Item 10

.634**

Item 11

.442* .824** .498* .688**

We also have studied the existence of correlations

among items using the Kendall tau-b correlation co-

efﬁcient, a non-parametric measure association as the

variables do not follow a normal distribution. We

have obtained some interesting ﬁndings, shown in Ta-

ble 2. The positive correlations between item 1 and

item 11 reﬂect that fun and motivation are closely re-

lated factors. The correlation between item 2 and item

5 shows that the greater the interest in the activity,

the greater the concentration of girls on their perfor-

mance. Correlations between item 10 and items 2,

6 and 9 indicate that the idea of having this system

at home makes them be more motivated to carry out

rehabilitation activities. On the other hand, the posi-

tive correlations among item 11 and items 14, 15, 16

and 17, indicate that interface elements contribute to

motivate the girls to do their rehabilitation exercises.

The absence of correlation with the item related to the

avatar (item 13) can give us some extra information in

the sense that its appearance should be adapted to the

user. Finally, the positive correlations among item 17

and items 2, 6, 10 and 11 seem to indicate that the in-

terface contributes to capture the user’s interest. All

of these correlations can be observed graphically in

Figure 4, grouped into three categories: activity per-

ception, usefulness, interface.

4.4 Limitations

This pilot experience presents several threats to inter-

nal and external validity (Shadish et al., 2002) that

might have inﬂuenced our results.

• Construct Validity. Some items (5, 7, and 8) were

presented in a negative way respecting the others,

what has led to a confusion in the answers. This

should be taken into account in future experiments

to avoid erroneous answers.

• Statistical Conclusion Validity. Given the limited

power of the sample because of its size and its

domain, our data exploration has consisted in a

statistical description and a basic correlation anal-

ysis among items. A deeper study should be per-

formed in future experiments in order to contrast

answers provided by different groups of partici-

pants.

• Internal Validity. Although quasi-experiments

avoid most of the threats to internal validity that

arise in other kind of experiments (B

arnighausen

et al., 2017), one limitation is related to the in-

terpretation of the results, since it is necessary to

consider the possibility that they can be inﬂuenced

by other factors not taken into account (Cook and

Campbell, 1986). For example, the 5 value has

been the most frequent one assigned in the ques-

tionnaire, which seems to indicate that girls, espe-

cially the youngest, were not very clear about the

difference between the range of values. Further-

more, since rhythmic gymnastics is a sport that

demands a high level of perfection, girls may be

inﬂuenced, unconsciously, by this level and have

thought that the only value which reﬂects a posi-

tive evaluation was the highest. This factor should

be taken into account in subsequent research ex-

periments.

• External Validity. The sample is not representa-

tive for the general population. Besides, data col-

lected in a gymnastics class may not be general-

ized to other educational scenarios as they are in-

evitably subjected to bias and non-bias systematic

experimental measurement errors. Moreover, al-

though we have considered a generic research hy-

pothesis, its empirical approach has forced us to

focus on a concrete context and with a speciﬁc

tool. Therefore, the replication of this study in

other contexts and with other users remains open

as an important future working line.

ICEIS 2020 - 22nd International Conference on Enterprise Information Systems

540

Figure 4: Correlation between variables grouped into activity perception, usefulness and interface.

5 CONCLUSIONS

In this article a system is designed to rehabilitate chil-

dren and teenagers suffering from lower back pain.

The rehabilitation tasks are customized by a high level

language called PEL, based on JSON standard, that

extends in the glTF speciﬁcation. The objective of

the language in this research is aimed at rehabilitat-

ing patients with bone-marrow injuries, although the

nature of the language makes it possible be extrapo-

lated to other rehabilitation contexts. The exergames

are ﬁrst designed by therapists, who deﬁne the ther-

apeutic objective, the trajectory of the rehabilitation

movement and the metrics to measure the patient’s

progress, all based on his or her conditions. Then, de-

velopers translate this information into PEL sentences

for automatically generating the exergames by means

of parsing process.

This article also includes a preliminary experi-

ment with potential patients to evaluate the system in

terms of understanding and suitability. To do this, an

exergame was developed to motivate the potential pa-

tients to perform exercises aimed at the physical reha-

bilitation of the lumbar spine. From a general point

of view, most participants perceived the tool as an ex-

cellent starting point to facilitate the process of patient

rehabilitation, considering it fun, interesting, and easy

to use.

As future lines of research, we can stress the need

to work on two core topics: (1) developing a system

or module that is capable of informing the therapist,

by way of notiﬁcations, on how children or teenagers

are progressing, and (2) constructing a visual tool to

help therapists graphically set the parameters, which

deﬁne the dynamics of an exergame, translating the

visual deﬁnition into PEL sentences.

ACKNOWLEDGEMENTS

This research was partially funded by Instituto

de Salud Carlos III, grant number DTS18/00122,

co-funded by the European Regional Development

Fund/European Social Fund “Investing in your fu-

ture”, by the Department of Technologies and Infor-

mation Systems (grant number 00421372), and by

the University of Castilla-La Mancha (AIR Research

Group, grant number 01110G9180).

REFERENCES

arnighausen, T., Tugwell, P., Røttingen, J.-A., Shemilt,

I., Rockers, P., Geldsetzer, P., Lavis, J., Grimshaw,

J., Daniels, K., Brown, A., et al. (2017). Quasi-

Personalising Exergames for the Physical Rehabilitation of Children Affected by Spine Pain

541

experimental study designs series—paper 4: uses and

value. Journal of clinical epidemiology, 89:21–29.

Clark, R. A., Pua, Y.-H., Fortin, K., Ritchie, C., Webster,

K. E., Denehy, L., and Bryant, A. L. (2012). Valid-

ity of the microsoft kinect for assessment of postural

control. Gait & posture, 36(3):372–377.

Cook, T. D. and Campbell, D. T. (1986). The causal as-

sumptions of quasi-experimental practice: The origins

of quasi-experimental practice. Synthese, pages 141–

180.

Da Gama, A., Fallavollita, P., Teichrieb, V., and Navab, N.

(2015). Motor rehabilitation using kinect: a system-

atic review. Games for health journal, 4(2):123–135.

Davis, F. D. (1993). User acceptance of information tech-

nology: system characteristics, user perceptions and

behavioral impacts. International journal of man-

machine studies, 38(3):475–487.

Deterding, S., Khaled, R., Nacke, L. E., and Dixon, D.

(2011). Gamiﬁcation: Toward a deﬁnition. In CHI

2011 gamiﬁcation workshop proceedings, volume 12.

Vancouver BC, Canada.

Deutsch, J. E., Robbins, D., Morrison, J., and Bowlby, P. G.

(2009). Wii-based compared to standard of care bal-

ance and mobility rehabilitation for two individuals

post-stroke. In 2009 virtual rehabilitation interna-

tional conference, pages 117–120. Ieee.

Duthey, B. (2013). Background paper 6.24 low back pain.

Priority medicines for Europe and the world. Global

Burden of Disease (2010),(March), pages 1–29.

Esculier, J.-F., Vaudrin, J., Beriault, P., Gagnon, K., and

Tremblay, L. E. (2012). Home-based balance train-

ing programme using wii ﬁt with balance board for

parkinson’s disease: a pilot study. Journal of Rehabil-

itation Medicine, 44(2):144–150.

Fernandez-Cervantes, V., Neubauer, N., Hunter, B., Strou-

lia, E., and Liu, L. (2018). Virtualgym: A kinect-

based system for seniors exercising at home. Enter-

tainment Computing, 27:60–72.

Gonz

alez, C. S., Toledo, P., Padr

on, M., Santos, E., and

Cairos, M. (2013). Tango: H: creating active educa-

tional games for hospitalized children. In Manage-

ment Intelligent Systems, pages 135–142. Springer.

Gonz

alez, C. S. G., del R

ıo, N. G., and Adelantado, V. N.

(2018). Exploring the beneﬁts of using gamiﬁcation

and videogames for physical exercise: a review of

state of art. IJIMAI, 5(2):46–52.

Hardy, S., Dutz, T., Wiemeyer, J., G

obel, S., and Steinmetz,

R. (2015). Framework for personalized and adaptive

game-based training programs in health sport. Multi-

media Tools and Applications, 74(14):5289–5311.

Hoy, D., Brooks, P., Blyth, F., and Buchbinder, R. (2010).

The epidemiology of low back pain. Best practice &

research Clinical rheumatology, 24(6):769–781.

Jones, M., Stratton, G., Reilly, T., and Unnithan, V. (2004).

A school-based survey of recurrent non-speciﬁc low-

back pain prevalence and consequences in children.

Health education research, 19(3):284–289.

Katajapuu, N., Luimula, M., Theng, Y. L., Pham, T. P.,

Li, J., Pyae, A., and Sato, K. (2017). Beneﬁts of

exergame exercise on physical functioning of elderly

people. In 2017 8th IEEE International Conference on

Cognitive Infocommunications (CogInfoCom), pages

000085–000090. IEEE.

Lai, C.-L., Huang, Y.-L., Liao, T.-K., Tseng, C.-M., Chen,

Y.-F., and Erdenetsogt, D. (2015). A microsoft kinect-

based virtual rehabilitation system to train balance

ability for stroke patients. In 2015 International Con-

ference on Cyberworlds (CW), pages 54–60. IEEE.

Matallaoui, A., Koivisto, J., Hamari, J., and Zarnekow, R.

(2017). How effective is “exergamiﬁcation”? a sys-

tematic review on the effectiveness of gamiﬁcation

features in exergames. In Proceedings of the 50th

Hawaii International Conference on System Sciences.

McCallum, S. (2012). Gamiﬁcation and serious games for

personalized health. In pHealth, pages 85–96. IOS

Press.

McCambridge, J., Witton, J., and Elbourne, D. R. (2014).

Systematic review of the hawthorne effect: new con-

cepts are needed to study research participation ef-

fects. Journal of clinical epidemiology, 67(3):267–

277.

Mobini, A., Behzadipour, S., and Saadat Foumani, M.

(2014). Accuracy of kinect’s skeleton tracking for up-

per body rehabilitation applications. Disability and

Rehabilitation: Assistive Technology, 9(4):344–352.

Mousavi Hondori, H. and Khademi, M. (2014). A review

on technical and clinical impact of microsoft kinect on

physical therapy and rehabilitation. Journal of medi-

cal engineering, 2014.

Murray, C. J., Vos, T., Lozano, R., Naghavi, M., Flax-

man, A. D., Michaud, C., Ezzati, M., Shibuya, K.,

Salomon, J. A., Abdalla, S., et al. (2012). Disability-

adjusted life years (dalys) for 291 diseases and injuries

in 21 regions, 1990–2010: a systematic analysis for

the global burden of disease study 2010. The lancet,

380(9859):2197–2223.

O’Sullivan, S. B., Schmitz, T. J., and Fulk, G. (2019). Phys-

ical rehabilitation. FA Davis.

Palacios-Navarro, G., Garc

ıa-Magari

no, I., and Ramos-

Lorente, P. (2015). A kinect-based system for lower

limb rehabilitation in parkinson’s disease patients: a

pilot study. Journal of medical systems, 39(9):103.

Pirovano, M., Surer, E., Mainetti, R., Lanzi, P. L., and

Borghese, N. A. (2016). Exergaming and rehabilita-

tion: A methodology for the design of effective and

safe therapeutic exergames. Entertainment Comput-

ing, 14:55–65.

Robinet, F., Arnaud, R., Parisi, T., and Cozzi, P. (2014).

gltf: Designing an open-standard runtime asset for-

mat. GPU Pro, 5:375–392.

Shadish, W. R., Cook, T. D., Campbell, D. T., et al. (2002).

Experimental and quasi-experimental designs for gen-

eralized causal inference/William R. Shedish, Thomas

D. Cook, Donald T. Campbell. Boston: Houghton

Mifﬂin,.

Shull, F., Singer, J., and Sjøberg, D. I. (2007). Guide to

advanced empirical software engineering. Springer.

Sparks, D., Coughlin, L., and Chase, D. (2011). Did too

much wii cause your patient’s injury? The Journal of

family PracTice, 60(7).

ICEIS 2020 - 22nd International Conference on Enterprise Information Systems

542

Sweller, J., Van Merrienboer, J. J., and Paas, F. G. (1998).

Cognitive architecture and instructional design. Edu-

cational psychology review, 10(3):251–296.

Webster, D. and Celik, O. (2014). Systematic review of

kinect applications in elderly care and stroke rehabil-

itation. Journal of neuroengineering and rehabilita-

tion, 11(1):108.

Wiemeyer, J., Deutsch, J., Malone, L. A., Rowland, J. L.,

Swartz, M. C., Xiong, J., and Zhang, F. F. (2015). Rec-

ommendations for the optimal design of exergame in-

terventions for persons with disabilities: Challenges,

best practices, and future research. Games for health

journal, 4(1):58–62.

Personalising Exergames for the Physical Rehabilitation of Children Affected by Spine Pain

543