Entertainment Computing 14 (2016) 55–65

Contents lists available at ScienceDirect

Entertainment Computing
journal homepage: ees.elsevier.com/entcom

Exergaming and rehabilitation: A methodology for the design of effective

and safe therapeutic exergames q
Michele Pirovano a, Elif Surer a,1, Renato Mainetti a, Pier Luca Lanzi b, N. Alberto Borghese a,⇑
a
Laboratory of Applied Intelligence Systems, Department of Computer Science, Università degli Studi di Milano, Via Celoria 20, 20133 Milan, Italy
b
Dipartimento di Elettronica, Informatica e Bioingegneria, Politecnico di Milano, Milano, Italy

a r t i c l e i n f o a b s t r a c t

Article history: We present here a comprehensive definition of therapeutic exergames from which a methodology to cre-
Received 4 February 2015 ate safe exergames for real therapy pathways is derived. Three main steps are identified. (I) A clear iden-
Revised 6 October 2015 tification of all the exercise requirements, not only in terms of goals of the therapy, but also in terms of
Accepted 26 October 2015
additional constraints. Characteristic parameters for determining the challenge level and to assess pro-
Available online 31 October 2015
gression are also defined in this phase. (II) The exercise is transformed into a Virtual Exercise, in which
all the exercise elements are implemented inside a simple virtual environment. In this step the discussion
Keywords:
between clinical and ICT teams allows maximizing the effectiveness of exergames implementation. (III)
Exergames
Serious games
The final exergame is realized by introducing on top of the exercise all the game elements suggested by
Game therapy good game design to maximize entertainment. A clear line between exercises and games is drawn here.
Rehabilitation We illustrate the methodology with exergames designed for (1) balance and posture and (2) neglect reha-
Game design bilitation, implemented and tested with post-stroke patients training autonomously at home. The
methodology can have a broader impact as it can be applied also in other gaming fields in which the
requirements go beyond entertainment.
Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction Exergames have become widespread in the well-being field,

Physical rehabilitation is nowadays ubiquitous and keeps
expanding as the elderly population increases in industrialized
countries. Stroke and traumatic brain injuries (TBI) represent the
leading medical conditions that require intensive rehabilitation
[1]. In the traditional healthcare model, rehabilitation consists of
daily sessions of exercises carried out with a therapist supervising
the patient in real-time [2]. This represents a large and ever-
increasing cost for healthcare providers that cannot be sustained
in the long term, hence why new solutions to administer rehabili-
tation are needed. In this context, a lot of research has recently
been directed toward video games, and exergames in particular.
Exergames let patients exercise while playing games that hide
the burden of the therapeutic repetitive tasks under the hood of
a compelling fantasy, thus providing effective treatment while
leveraging the motivational power of games to increase adherence.
q
This paper has been recommended for acceptance by Helmut Hlavacs.
⇑ Corresponding author. Tel.: +39 02 503 14011.
E-mail address: alberto.borghese@unimi.it (N. Alberto Borghese).
1
Present address: Department of Modeling and Simulation, Graduate School of
Informatics, Middle East Technical University, Ankara, 06800, Turkey.
even finding commercial success, like the Wii Fit game.2 An
increasing number of recent studies have introduced exergames in
rehabilitation [3,4]; these demonstrate that exergame-based therapy
is effective and that motivational benefits do exist. However, current
approaches require the presence of a therapist, at least remotely, for
safety and efficacy reasons. This has limited so far the adoption of
this approach in real settings as the gain appears small.
A few attempts were made in using commercial exergames for
autonomous rehabilitation at home [5]. However, adverse effects
were reported, such as occurrences of knee or back pain, and too
little validation data are available as of now to conclude on their
beneficial properties. Similar problems were reported also for the
use of the Wii for fitness purposes [6].
Indeed, commercial exergames, made for entertainment and fit-
ness, cannot address the whole range of requirements of a rehabil-
itation therapy. In fact, commercial exergames integrate into their
gameplay the primary goals of an exercise, which are translated
into game actions, but they do not address what we call here the
secondary goals of an exercise, i.e. the specification of how an exer-
cise should be carried out. This is important, for instance, in reha-
2
Available at http://wiifit.com/.

http://dx.doi.org/10.1016/j.entcom.2015.10.002
1875-9521/Ó 2015 Elsevier B.V. All rights reserved.
56 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65
bilitation, where therapists not only assign exercises, but also
supervise, advise, and correct the patient while exercising. With-
out this supervision activity, exercising could possibly do more
harm than good due to maladaptation, wrong postures, and joint
overloading [5,6].
Exergames can represent the key to provide autonomous reha-
bilitation at-home, but they should combine motivational content
with supervision to guarantee therapy efficacy and safety. This
raises a question: how can we design such exergames to guarantee
this? Some initial work in this direction was carried out by Flores
et al. [7] which pointed out the lack of common guidelines and
made some initial proposals, but a shared view of which features
a therapeutic exergame should possess is still missing. The discus-
sion is also made difficult by confusion about terminology, a con-
fusion that must be lifted if we want to create effective
therapeutic exergames. In this paper, we aim to fill this gap. We
clearly define what exergames and therapeutic exergames are,
and, based on this, we outline a methodology for the design of
exergames of therapeutic validity. We also show examples of
use, and discuss the consequences of our approach applied to the
rehabilitation domain.
2. Materials and methods
2.1. Defining exergames and therapeutic exergames
Although the term exergame has become widespread, its defini-
tion is still fuzzy. An exergame has a dual nature: it is both an exer-
cise and a game, but it is hard to separate the two aspects.
However, by drawing a line between these two aspects, the task
of designing the exercise, its graphical appearance, or the gaming
elements can be better approached by the respective field experts.
Recently, Oh and Yang reviewed the use and definition of exergame
in research, addressing the characteristics of the exercise aspects
and proposing their own all-encompassing definition: ‘‘an experi-
ential activity in which playing an exergame or a video game requires
physical exertion or movements that are more than sedentary activi-
ties and also include strength, balance, and flexibility activities” [8].
However, neither this definition analyses the interplay between
the game and the exercise aspects nor does it neatly separate the
two. Moreover, current exergames are focused on what a user
should do (primary goals) and totally disregard how user actions
should be carried out (secondary goals). In addition, previous def-
initions do not give any insight on how to design exergames. Based
on these observations, we provide here a novel comprehensive def-
inition of exergames and we also make a clear distinction with
exergames that can be applied in real clinical settings, referred to
as therapeutic exergames.
We start from the definition of exercise. The concept of exercis-
ing is well known and leaves little doubt to its interpretation. We
borrow its definition from a medical dictionary [9]:
1. ‘‘Exercise is physical activity that is planned, structured, and
repetitive for the purpose of conditioning any part of the body.”
From this definition, it is clear that exergames can be fully con-
sidered exercises. Finding a suitable definition of game is harder, as
even if many authors have endeavored to find an all-encompassing
definition of what a game is, a consensus has not been reached yet.
Dictionaries tend to define a game as ‘‘a physical or mental contest,
played according to specific rules, with the goal of amusing or reward-
ing the participants” [10]. However, different interpretations of the
term do exist depending on the author’s focus, and the fact that the
term game has many different uses makes a common definition
even harder to achieve. Salen and Zimmermann [11] performed a
throughout comparison of the definitions in the literature and
provided a popular definition that includes most of the commonly
accepted elements:
2. ‘‘A game is a system in which players engage in an artificial
conflict, defined by rules, that results in a quantifiable outcome.”
These definitions have a point in common: an exercise is struc-
tured, while a game is defined by rules. The two concepts can be
assimilated. Of the two, especially while treating therapeutic exer-
games, the exercise’s structure must clearly have priority, as our
main goal must be to provide valid exercises, and only then to pro-
vide entertaining games. By defining how the exercise and the
game are linked through this common point, we will come up with
a suitable definition of exergame. Given a game, the person that
interacts with it is said to be playing. Much like game, play is a term
that has many different meanings. We borrow, again from [11], a
quite general definition of play:
3. ‘‘Play is free movement within a more rigid structure.”
This definition better clarifies the relationship between a game
and an exercise: since an exercise is defined through its structure,
we derive that by introducing play into the exercise we manage to
assemble a game inside the exercise itself. This last observation
provides a natural definition for the term exergame:
4. ‘‘An exergame is an exercise with a game built into its structure.”
Compared to other definitions, this definition details the rela-
tionship between exercise and game, as it highlights the primacy
of the exercise over the game. It follows that the game should
not interfere with the correct execution of the exercise: this means
that the exergame, when stripped of its gaming parts, should still
work as a valid exercise.
This has a large impact on the game designer as game mechan-
ics must be constrained by the exercise limits. The freedom of
movement needed for the player to enjoy the game (its gameplay)
must be designed such that it is contained inside the structure of
the exercise, as visually suggested in Fig. 1. This constraint is even
stricter than may seem at a first glance, because not only does the
exercise constitute the structure for play, but it also dictates the
allowed physical movements that the user may perform and their
intensity. Game mechanics cannot modify such movements, as
they are required by the exercise.
By understanding the relationship between the game and the
exercise, we can now correctly treat primary and secondary goals.
In a therapeutic exergame, primary goals can be easily merged into
the gameplay, while secondary goals, such as movement correct-
ness or compensatory motion prevention, can be addressed sepa-
rately, although they should still provide feedback to the user.
From these considerations, we propose here a more complete def-
inition of therapeutic exergames:
5. ‘‘A therapeutic exergame is an exergame that supports all pri-
mary and secondary goals defined for an exercise.”
Only if all the aspects of an exercise are fulfilled a therapeutic
exergame can be considered really effective and safe for the patient.
2.2. A methodology for the design of therapeutic exergames
We show here how the prior definitions can be translated into a
suitable methodology to design effective and safe therapeutic exer-
games. This follows a four-step procedure (Fig. 2):
– Exercise definition. Starting from a therapy goal, a set of coher-
ent exercises that covers all the needs of the therapy is chosen.
Each exercise is properly structured in terms of primary and
secondary goals.
 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65 57

Fig. 1. On the left, a simple exergame is represented as an exercise (the outer structure) and a game (the inner circle). Intuitively, play can only be performed inside the
exercise’s structure that provides the game’s boundaries. On the right, a therapeutic exergame instead has its exercise separated into its primary and secondary goals. The
primary goals are associated to the gameplay, while secondary goals can be handled separately, for instance by a Virtual Therapist. Both give output feedback to the patient.
Notice that secondary goals may provide additional hints to steer exergame design, as discussed in the next section.

Exercise

1
Lateral Do not
Weight Primary Secondary Bend
Shi Elbows
2

Virtual Exercise

3 4

Exergame Monitoring

Therapeuc
Exergame

Fig. 2. A visual representation of the methodology to design therapeutic exergames. (1) The exercise is defined. (2) A virtual exercise is created to address primary goals while
keeping in mind secondary goals. (3) The VE is transformed into an exergame, merging game mechanics with the primary goals. (4) The secondary goals are addressed
through a monitoring system that is then integrated inside the exergame (5) to obtain therapeutic validity.

– Virtualization. The primary goals of the exercise are imple-
mented into a virtual exercise (VE) by defining input (tracking)
and output (feedback) requirements through simple graphical
elements and by specifying interaction mechanisms, while tak-
ing into account constraints on feedback relative to secondary
goals.
– Game design. The VE is then transformed into a true exergame
by adding all gaming elements.
– Secondary goals. The secondary goals of the exercise are han-
dled separately, finally achieving a therapeutic exergame.
2.2.1. Exercise definition
We start from the therapy goal of the exercise and we use reha-
bilitation as an example of application domain. The rehabilitation
domain over which a coherent set of exercises can be created has
to be first defined, typically arm, hand, or postural rehabilitation.
We then define the specific skills we want to re-train with each
exercise, for example, for posture rehabilitation, balance, muscle
strength, or shift of body support from one foot to the other. The
exercises must then be defined in terms of primary goals: the types
of movement elicited and its quality.
In this stage, language used by therapists should be adopted
and the exercise can be described as the temporal sequence of dif-
ferent actions needed to complete the exercise, such as doing lat-
eral steps, alternating left and right steps, or shifting the body
sequentially along random directions. This action sequence is then
broken down into constituent basic actions, each of which requires
a specific movement to be completed. We remark here that each
58 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65
basic action can be evaluated by itself. This allows us to distin-
guish, for instance, the capability of moving to the right versus
moving to the left.
For each exercise, quality parameters have also to be provided
to define movement properties. These can be, for instance, ampli-
tude, accuracy, or speed of motion, and they determine the degree
of challenge of the exercise itself and also allow an adaptation of
the exercise difficulty to patient needs. Lastly, a set of output
parameters has to be defined, for example movement accuracy,
success rate, or reaction time, which allows us to obtain quantita-
tive evaluation of the patient’s performance on the primary goals
of the exercise.
Even though this information is sufficient to build exergames, it
comes short in assigning therapeutic validity to them. In fact, noth-
ing stops patients from executing the action sequence with wrong
movements if unconstrained, stressing the unimpaired limb more
than the other. This can easily lead to anomalous compensatory
movements and to twisted postures, making the exercise useless
or even dangerous [5,6,12,13]. To guarantee safety, the exercise
definition should also clearly state secondary goals that define
how the movements are correctly executed. This information
may be defined in terms of allowed movements, such as ‘‘keep
the back straight”, ‘‘do not bend the knee”, and ‘‘keep the center
of mass between the feet”. Only by adding this information the def-
inition of a therapeutic exergame can be considered complete.
Both primary and secondary goals constitute the definition of
the exercise structure. To achieve maximum effectiveness, the def-
inition of the exercises should be carried out in strict collaboration
with the therapists [13] who identify these goals and clearly define
them, so that the final description of the exercises can be used as a
technical document for implementation.
2.2.2. Virtual exercises: Addressing primary goals
The now completely structured exercise needs to be trans-
formed into its virtual exercise counterpart (VE). A VE is concerned
with the primary goals of the exercise and addresses the exercise’s
required actions in terms of a sequence of targets to be reached
(Fig. 3). At the same time, it deals with basic choices related to
the virtual environment, such as choosing the point of view of
the user. In case of a third person view, the best angle from which
the avatar is seen from is also defined. Feedback mechanisms are
designed to clearly show the outcome of their actions to patients,
for instance whether a target is hit or a movement has been suc-
cessfully performed (Fig. 3a). We also define tracking require-
ments: the most adequate set of input devices that track all the
required movements are identified. Since these choices impact
on the feedback related to secondary goals (Section 2.2.4), good
Fig. 3. A virtual exercise associated to a 360 Steps exercise (Section 3.1.1) is shown. A third person view is used: the user sees his avatar inside an empty environment in which
the only elements, besides the avatar, are red spheres representing targets that appear, one at a time, at a set distance around the patient’s avatar (a) and disappear when the
user steps over them with a green sign and an auditory feedback (b). (For interpretation of the references to color in this figure legend, the reader is referred to the web
version of this article.)
design practice requires that these needs are taken into account
in this design stage. For example, we may choose a humanoid ava-
tar instead of simpler objects if we want to give feedback on full-
body posture. This choice forces the final exergame to adopt this
player representation. Therefore, secondary goals may produce
additional constraints to the design of feedback mechanisms (cf.
Fig. 1b) and the VE is the place where these constraints are taken
into account.
The VE requires a rendering engine, code implementing the
interaction logic, bare graphical elements that work as targets
and feedbacks, the required input devices, and an output device
(typically a display). The VE should be configurable so that input
parameters associated to game difficulty can be changed and their
effect analyzed. As a result of this phase, an exercise can be per-
formed inside a virtual environment without the distractions of
the gaming enrichments. The VE is the common ground on which
the development team can analyze the exergame with the clini-
cians to evaluate it early on, before full development, following
an iterative prototyping design method. This could be also
regarded as the game’s skeleton, the result of stripping away all
graphical elements to expose the underlying basic actions, making
it quite similar to many very simple games proposed early in the
rehabilitation domain [14,15]. Additional features to support the
therapy can be added to a VE upon need, such as dynamic difficulty
adaptation [16] and data logging.
2.2.3. Game design for exergames
The next step is to transform a VE into an enjoyable exergame
by introducing gaming characteristics developed according to the
rules of good game design [11,17,18], and especially taking into
account good game design for elderly patients [19] and for rehabil-
itation exergames in particular [7,13]. The game components
should be designed such as to enforce the underlying exercise
and put accessory elements in the background. In order to design
a compelling game constrained inside an exercise, we can work
on the gaming elements that are not strictly related to game
actions and mechanics. As a basis, all games require meaningful
play (i.e. that actions have meaningful consequences in the game),
clear and immediate feedback, as well as simple and direct interac-
tions. These principles should be geared toward making the under-
neath exercise clear and intuitive to perform. The exergame
motivational factor can be further increased by leveraging intrinsic
‘‘fun factors”: challenge [20], fantasy [21], curiosity, sensation, and
social play. These can be placed on top of the bare exergame, where
possible. Fantasy and sensation are especially easy to introduce
without affecting the exercise. At last, extrinsic motivational
effects can be obtained through careful use of verbal praise, scoring
 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65
mechanisms, and virtual reward systems [22,23]. The latter ele-
ments, being extrinsic, do not modify the exercise at all and are
thus easy to introduce. As a final result, a VE is transformed into
an exergame, with the motivational benefits this entails.
Finally, early testing with healthy subjects and then with
patients should be carried out to evaluate compliance and usability
and to receive feedback to improve the game design. This is also
done in [24], where it is shown that good game design and graphics
largely improve the enjoyment provided by an exergame, although
highly detailed graphics do not add any significant advantage com-
pared to simpler, but refined, graphics.
2.2.4. Handling secondary goals
To handle secondary goals, we must design two different func-
tionalities: the first analyzes the stream of motion data and identi-
fies in real-time wrong movements, the second provides an
immediate feedback to the patients who can correct themselves
while exercising. An example of effective implementation of a
monitoring system is described in [25] where a fuzzy monitoring
system is employed to supervise patient motion. Feedback on
wrong movements is provided by changing the color of the associ-
ated avatar segments, from green (correct movement), to yellow,
orange, and red, depending on severity. When very wrong move-
ments persist for a long time, the game is paused and a Virtual
Therapist Avatar pops up to advise patients. Such a system there-
fore guarantees safety to the patient and constitutes a sort of con-
tinuity between hospital and home rehabilitation and it is
therefore highly important for compliance.
Monitoring functionalities can be placed above the gameplay, as
they are independent of it: their implementation neither modifies
the game, nor is it integrated in its fantasy, the gameplay basically
ignores it. Therefore, monitoring functionalities can be switched on
or off by the therapist without hampering the use of the exergame.
Even though game and monitoring are independent, they share the
same input tracking and output feedback mechanisms, which is
beneficial for reuse. Notice that monitoring may very well reuse
the same tracking devices that were selected for gameplay (i.e. to
track the action sequence). Feedback may use the avatar’s appear-
ance, but also non-game related visual and audio feedback, such as
generic color-coded bars or simple and direct sounds.
3. Results
We describe here how this methodology can be used to design
therapeutic exergames for rehabilitation in two different domains:
balance and posture rehabilitation and neglect rehabilitation.
3.1. Balance and posture exergames
We must first identify the skills on which the patient has to be
trained. These are: balance, strength, and endurance [13]. The
patient first needs to re-learn to acquire a vertical posture, with
his/her center of mass centered between the feet. He/she then
can move toward exercises that require to shift the body, first lat-
erally, then in the antero-posterior direction, and finally in all
directions. Finally, the patient must relearn to step laterally, for-
ward/backward, and in all directions. In parallel, he/she has to
increase the power of ankle and knee extensor muscles. Therapists
must then identify typical exercises to improve these skills.
As an example, we will describe completely three exercises:
Lateral Weight Shift, which requires patients to shift their weight
from the left to the right leg (and vice versa) without changing
the feet position, and Steps, which requires patients to perform
steps in different directions. Two different versions of the latter
exercise are considered: Lateral Steps and 360 Steps. The first
59
Table 1
The input parameters used in the definition of the balance and posture exercises.
Name Description Unit
Area of movement Required movement range m
Spatial accuracy Accuracy needed to consider the interaction m
successful
Trial duration Time interval between the start of a trial and s
its end
Inter-trial duration Time interval between one trial and the next s
Touch time Time elapsed at the target position to consider s
the interaction valid
Total duration Total duration of the exercise S
Cognitive load For instance distractors or color coding n/a
Table 2
The output parameters used to assess the performance in the balance and posture
exercises.
Name Description Unit
Reaction time Time elapsed from trial appearance to completion s
Maximum Maximum achieved movement (in four directions) m
movement
Accuracy Achieved accuracy when completing a trial m
requires only lateral steps, while the second guides the patients
through steps over 360 degrees. These two will be treated differ-
ently in the virtualization step.
We consider three fundamental characteristics for these exer-
cises: range, accuracy, and speed, keeping in mind that in rehabil-
itation a patient is first trained on wider movements and in a
second stage on more accurate and faster movements [26]. From
these considerations we derived the quality parameters that regu-
late these characteristics: area of movement, spatial accuracy, and
duration of each basic action. Additional parameters can be
defined, namely the touch time and the inter-trial period. The total
duration of an exercise also has to be defined, either in terms of
number of trials or total time. Lastly, in some cases, a cognitive
load value can be added to the parameters, adding the requirement
of making forced choices while exercising. For instance, only tar-
gets of a given color or shape are valid and have to be reached.
These parameters are summarized in Table 1. We remark that
the area of movement parameter has a different definition for
the different exercises: in Lateral Weight Shift it is related to the lat-
eral Limits of Stability (LOS)3 of the patient, while in Steps exercises
it is related to the maximum step length of the patient.
We also define with the therapists the parameters used for eval-
uating patient performance (Table 2).
In addition to these primary goals, therapists have to define
how these exercises should be executed to avoid maladaptation.
In the case of Lateral Weight Shift and Lateral Steps, knees and spine
should not be bent and the center of pressure should not be moved
orthogonally to the direction of movement. For the 360 Steps exer-
cise, only the knee and spine constraints are applied.
Evaluation of the efficacy of therapeutic exergames has to be
carried out through clinical assessment administered to the patient
before and after the treatment. These tests can also be adminis-
tered in between to tune the therapy adapting the difficulty level
and the parameters of the games. In the postural domain, for
instance, the following classical assessments have been suggested
[13]: Berg Balance Berg Balance Scale, 7-min Timed Up and Go,
Short Physical Performance Battery, force platform stance tests,
3
LOS quantifies the maximum distance a person can intentionally displace their
Center of Gravity (COG), i.e. lean their body in a given direction without losing
balance, stepping, or reaching for assistance.
60 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65
Fig. 4. The Fruit Catcher – Lateral Steps exergame (a), the Fire Fighter – 360 Steps exergame (b), and the Bubbles Bursters – 360 Steps exergame (c).
and gait analysis. These assessment tests are clearly linked to the
primary goals defined for the exergames.
3.1.1. Virtual exercises for balance and posture
To virtualize the exercises, we created an essential virtual envi-
ronment: targets are represented as colored spheres and the user is
represented as an avatar. No other element is added at this stage
(Fig. 3). The user is instructed to perform the required movements,
e.g. stepping laterally, and upon reaching the target it disappears,
an audio signal is played, and a tick is displayed besides it
(Fig. 3b). To match primary and secondary goals, i.e. to track users’
movement and to provide the data required to monitor them, input
requirements have to be defined. The selected exercises require
both the measurement of the patients’ center of pressure (COP)
and of their full body movement (especially to monitor secondary
goals). We thus chose the Nintendo Wii Balance Board to track the
COP, and the Microsoft Kinect sensor to track whole body move-
ment in three dimensions.
In all these VEs, a third-person viewpoint was chosen with ther-
apists, so that patients see themselves from behind for a more
effective feedback. In Lateral Weight Shift and Lateral Steps, as the
movement is on the sagittal plane, the camera is positioned fron-
tally. For 360 Steps, instead, we created two different VEs. In one,
the virtual camera is positioned slightly above to make the dis-
tance of each target clear in all directions but still allow being able
to use a humanoid avatar (Fig. 4a and b). In the other, the point of
view is from above, and the avatar is a graphical object, like the
wooden stick in Fig. 4c. The above view makes virtual distances
very clear on all eight directions and it is preferred for gameplay.
However, this view does not easily allow feedback on the patient’s
posture and step height required by monitoring. For this reason, it
is a worse choice when monitoring is required.
3.1.2. Therapeutic exergames for balance and posture
We built several exergames over the virtual exercises defined
above and we describe here how the defined methodology has
shaped three of them: Fruit Catcher over Lateral Weight Shift and
Lateral Steps, Fire Fighter over Lateral Steps and 360 Steps, and Bub-
bles Burster over 360 Steps.
In Fruit Catcher (Fig. 4a), the player takes the role of a farmer
that must pick apples falling from a tree to fill a basket. The apples
spawn on the tall tree and the player’s avatar cannot reach them
with its hands. Instead, the avatar has a basket on its head, which
can be used to collect the apples that fall before they touch the
ground. At each trial, a small apple appears on the tree, grows,
and then falls down. If an apple touches the ground, the trial is con-
sidered a failure. If the player manages instead to catch the apple
with the basket, the trial is considered a success. Movement range
determines here the lateral position from which the fruit can fall,
accuracy determines the basket amplitude, and speed is associated
to the falling time of each fruit (height of spawn). Virtual objects
inside the VE are transformed into nice cartoons, while additional
graphical objects are added to the scene to make the environment
richer and to reinforce the farming fantasy. These additional
objects are placed to the sides, outside the place of interaction,
so to not distract the user from the game. To add a surprise ele-
ment, once in a while a worm will fall out of an apple after it is suc-
cessfully caught, disappearing into the grass.
In Fire Fighter (Fig. 4b), the player sees him/herself as a huma-
noid avatar inside a barn, surrounded by hay bales. Due to the high
temperature, the hay may catch fire and small flames may appear
around the player. The player must reach the fire to extinguish it.
The game was designed to work with the Steps exercises, in which
the player has to step over the virtual flames to extinguish them. In
this game the range of motion determines the spawning radius of
the flames, the speed parameter determines the frequency with
which the flames appear, and the accuracy parameter determines
the amplitude of the flames.
In Bubbles Burster (Fig. 4c), the player sees him/herself as a stick
inside a big cauldron, seen from above. Bubbles appear on the sur-
face of the cauldron and must be reached by the stick by moving
around the play area. The game is built over the 360 Steps exercise.
The range parameters determine the spawning radius of the bub-
bles, the speed parameters determine the frequency with which
the bubbles appear, and the accuracy parameters determine the
size of the bubbles.
We remark that, to further reinforce the game fantasy through-
out all exergames, all the games we designed share the same
theme: the farm life. Farm-related funny sound effects and match-
ing music (with also the possibility for patients to choose their own
tracks) are also added to all of our games. A scoring system is also
employed and, at the end of each exergame, a virtual reward is
obtained.
3.1.3. Secondary goals
Secondary goals are handled through a monitoring module that
watches over features extracted from the tracked input data
stream. For each feature, constraints can be defined at configura-
tion time by the therapist and a monitor is created to track varia-
tions over time to control that they do not violate the constraints.
The monitoring module can, for instance, be implemented through
a fuzzy system that translates into numbers the qualitative evalu-
ation of the therapist (safe, not completely safe, mildly safe, not
safe) of given features (e.g. trunk or knee straightness, center of
pressure between the feet) [25]. Each monitor controls the color
of a given feedback element, associated to the monitored feature,
such as the back, feet, or knees of the avatar (cf. Fig. 2) changing
it from green4 to red according to the degree of infringement of
the associated constraint. The monitor system should be able to
4
For interpretation of color in Fig. 2, the reader is referred to the web version of
this article.
 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65
Fig. 5. The virtual exercise Exploration and Cognition. Green targets must be reached, while red distractors must be avoided. On the left, a green target is picked and a green
tick is shown as positive feedback. On the right, a red distractor is picked and a red cross is shown as negative feedback. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)
pause the exergame when gross deviations from desired movements
are detected repetitively. When the humanoid avatar is not avail-
able, as in Bubbles Burster, the player controlled object (such as the
stick in the cauldron) changes color according to the behavior of
the monitor. Given that this feedback mechanic is less clear, two tilt
indicators are also overlaid on the virtual environment, to the sides
of the screen, to provide feedback on the actual orientation of the
critical body segments (Fig. 4c).
3.2. Exergames for neglect rehabilitation
Visuospatial neglect is a behavioral syndrome, usually stem-
ming from the right brain damage following a stroke, characterized
by the inability to interact with the stimuli from the contra-lateral
space [27]. Rehabilitation of neglect is based on visual scanning
tasks that allow the patient to recognize also stimuli located inside
the neglected space. Very recently, novel ICT platforms have pro-
vided new ways to tackle neglect. An automatic version of classical
quantitative neglect assessment tests has been proposed for
mobile platforms [28]. Video-installations [29], based on standard
web-cam, have also shown large potentiality, being simple and
easy deployable at patients’ home. An even simpler setup has been
identified following the methodology here described.
The starting point is the definition of the exercises that are all
based on visual scanning tasks [30]: patients have to reach targets
that appear in the whole visual field including also the neglected
hemi-space.
Recently, dual-task performance has been found to play a cru-
cial role in neglect [31], according to which some patients develop
neglect only when cognitive load exceeds a given threshold. We
have introduced explicitly cognitive load inside the exercises,
requiring the patients to handle forced choices in terms of color
and/or shape of targets and, in some exercises, to make choices
according to memorized targets. We will illustrate visual scanning
exercises with these characteristics: a basic exploration exercise, an
exploration exercise with target repositioning, an exploration exercise
with cognitive load, and an exploration exercise with memory.
We remark here that these exercises have the same input
parameters of the exercises for posture and balance and therefore
Table 1 applies also to this domain. Here the movement range
parameter determines the amplitude of the space to be scanned,
which is clipped inside the neglected side, where clipping is pro-
gressively moved toward the end of the hemi-space according to
patient’s amelioration. Accuracy parameters determine the size
of the target, and speed parameters determine trial duration and
the time span of each target. Evaluation parameters are the same
as those reported in Table 2.
A common requirement to all exercises is to add a mechanism
that attracts the attention of the patients to the neglected targets
so that he/she can find the targets also inside the neglected space.
61
3.2.1. Virtual exercises
The exercises defined require the tracking of the hand of the
patient in real-time in 3D. We identified a low-cost haptic tracker,
the Novint Falcon5 that is able to track hand motion in 3D inside a
volume of 100  100  100 mm3. Moreover, it can provide a force
field and we can use this to guide the patient’s hand toward the
neglected area when targets are not reached in a certain amount
of time, thus effectively implementing the active guidance mecha-
nism required.
According to therapists, an orthographic frontal view was
deemed the most suitable for these exercises as patients can see
all the targets available without any metric distortion due to pro-
jection (Fig. 5). Different numbers of targets can be shown on
screen depending on patient skills.
For Basic Exploration, each trial is composed of a set of virtual
targets that are displayed on the screen: when reached they disap-
pear. The targets are located on the right, center, and left areas of
the screen depending on the difficulty level of the exercise. When
cognitive load is added, the exercise type changes to Exploration
and Cognition (Fig. 5), where patient has to reach only the targets
that have a given color/shape. Exploration and Target Repositioning
requires more work from the patient, who has not only to pick a
target, but also to drag the target to a specific position. It also
requires cognitive reasoning for the repositioning phases, such as
by making it harder to find the correct end position. For the Explo-
ration and Memory exercise, three different mnemonic tasks are
considered here: (i) counting the number of targets that appear
on the display within an exercise, (ii) respect an ordered picking
sequence indicated on the screen, and (iii) picking the targets
according to a given initial list.
3.2.2. Exergames for neglect
Balloon Popper, Bubbles Burster, Puzzle, and Mix the Soup are four
of the therapeutic exergames developed for the four neglect exer-
cises defined above.
In Balloon Popper the user is placed in front of a virtual forest
where several red balloons move around. During a Basic Exploration
exercise, the user interacts with balloons with a hand icon, driven
by his/her own hand, and pops them when they are touched
(Fig. 6a). The same exergame can be used for an Exploration and
Cognition exercise. In this case, some blue balloons are also added
(Fig. 6b) that constitute distractors and must not be hit to avoid
losing points.
In the Bubbles Burster exergame (Fig. 7a) the user is placed in
front of a cauldron that contains water with some bubbles on the
surface. The user has to smash bubbles with a stick whose position
is controlled by the player’s hand. Cognitive load can be added by
asking the user to smash only bubbles of a given color. In this case,
5
Available at http://home.novint.com/index.php/products/novintfalcon.
62 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65

Fig. 6. The Balloon Popper – Basic Exploration exergame (left): the user has to reach the red balloons floating around. In the Balloon Popper – Exploration and Cognition exergame
(right), the user has to reach red balloons and not the blue ones. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of
this article.)

Fig. 7. (a) The Bubbles Burster exergame with cognitive load. The white dragon shows the color of the bubbles to be smashed. (b) The Puzzle exergame. The board represents
the puzzle’s final image. The user has to select one piece at a time and bring it into the right place. (For interpretation of the references to color in this figure legend, the reader
is referred to the web version of this article.)

Fig. 8. The Mix the Soup exergame. (a) The recipe is displayed on the screen at game start. (b) It then disappears during the game and the user is expected to remember the
ingredients and the already added ingredients.

a dragon figure is placed on the screen, which changes color in
order to show to the user which bubbles to smash. If the patient
smashes wrong-colored bubbles, he/she loses points.
The Puzzle exergame (Fig. 7b) is built over the Exploration and
Target Repositioning exercise. The patient has to complete a puzzle:
the user sees a puzzle board in the center and puzzle pieces are
spread around it. Through a hand icon, the user can pick up the
pieces and place them inside the board. If the puzzle piece is placed
in a wrong position, it returns automatically to its original position,
a warning sound is played and the user loses points. Depending on
the exercise’s cognitive load, the puzzle board can be represented
as the final puzzle image (Fig. 7), the final puzzle image with the
puzzle pieces outlined on it, only the puzzle pieces’ outlines, or
an empty board represented as a white rectangle.
The Mix the Soup exergame (Fig. 8) is built upon the Exploration
and Memory exercise, the objective is to cook a vegetable soup. The
ingredients are displayed on the left side of the screen whereas the
recipe is displayed on the right side. Thus, the user is expected to
explore both the right and left sides of the screen and to put the
ingredients into a pot. The recipe is displayed until the user starts
picking-up the first item. Afterward, the user has to remember the
initial list and already added ingredients. If the user adds an ingre-
dient which is not on the recipe, or tries to exceed the number of
items of a specific ingredient, he/she loses points. This exergame
is also extended with an additional task: the user has to regulate
the fire periodically to avoid that it grows out of control, for which
he/she loses points.
3.2.3. Secondary goals
The common difficulties of neglect patients are: (i) initiating
search to the left side, (ii) restricting search to the right side, (iii)
being focused to peripheral stimuli on the right side, and (iv) being
hesitant to rescan an array once the array is viewed [32]. Therefore,
the acquisition of eye movements during the exercise can provide
 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65
further and more detailed insights on the patient’s actual status
and help patients to improve. Johnston and Diller [33] suggested
that feedback of eye movements could encourage patients to move
their eyes. Trepagnier [34] also mentioned that using displays that
provide feedback based on gaze locus may be particularly applica-
ble to computer-aided neglect training. All these observations sug-
gest that gaze tracking can be required by the system, besides
hands tracking for a complete monitoring. This could be done for
example through SensoMotoric Instruments’ gaze and eye tracking
systems, which are especially used in cognitive field research [35].
Head orientation is even more important. Patients know that
they are missing something on their left during rehabilitation ses-
sions, so they have the tendency to rotate their head toward the
neglected hemi-space to see stimuli. However, this kind of com-
pensation is not what is desired and should be avoided. To solve
this problem, real-time monitoring can be achieved through head
tracking easily provided by standard Web-cam advanced function-
alities: software such as FaceTrackNoir or Logitech’s tracking soft-
ware [36,37]. During gameplay a warning can be given to the
patient, similarly to the case of balance and posture, when he/
she rotates his/her head off the central position.
4. Discussion
In recent years, the development of new exergames for health
has been blooming, but their use in real therapeutic paths is still
limited. The lack of compliance with patients and the poor match
of all requirements associated with a therapy are the main reasons.
Most exergames are built or chosen according to the primary goals
of an exercise, i.e. to elicit a given movement. However, for therapy
in general, and for rehabilitation in particular, the way in which an
exercise is executed, here referred to as secondary goals, is equally
important. For example, we may require that, while gaming, the
patient keeps a symmetric distribution of the body weight on the
support, does not tilt the spine, and does not bend the knees. This
avoids an overloading or maladaptation that would make rehabil-
itation more harmful than useful. These secondary goals that con-
tribute to the safety and effectiveness of the exercise are neglected
in typical exergames, which are usually targeted to the fitness mar-
ket and therefore to healthy people (e.g. Wii Fit), making the use of
commercial exergames in rehabilitation unsafe [5,6,12,13].
Recently, the ski game of Wii Fit has been re-implemented to pro-
vide movement data acquisition for performance assessment [38].
Although modifying successful commercial exergames to add the
required features that make them suitable to rehabilitation is a
possible solution, starting with all primary and secondary goals
clear in mind is a more sound approach. This is a critical point that
has gone largely unnoticed and it goes beyond the common view of
exergames seen first as games, and then as exercises, while we
remark the primacy of a good and comprehensive exercise
definition.
From these observations, we came up with definition 5, in
which we stress the importance of both primary and secondary
goals. The methodology described in Section 2 takes this in full
account. It allows a clear identification of the roles and responsibil-
ities of the clinical and the ICT teams during therapeutic exergame
design.
A complete definition of the exercises required by the domain
expert is the starting point of the methodology. Such definition is
translated into simple mechanics inside the VE, as explained by
the examples in Section 3. A VE can be seen as a sort of ‘‘naked”
game (stripped away of all graphical enrichments and game goals)
that exposes the bare mechanics associated to the exercise (Fig. 2).
As such, the VE plays a central role, as it is the common ground on
which the two teams can interact to clearly identify exergame
63
aspects related to the therapy and tune any further development
of the prototype into a refined exergame. This step is fundamental
to guarantee that the final exergame meets all therapy require-
ments. For instance, constraints in the choice of feedback given
to the player can arise: in the initial rehabilitation phase, in which
the patient has to regain vertical posture, a clear view of the entire
skeleton is required and avatars like those shown in Fig. 4a–b are
enforced. In a more advanced therapy stage, a simplified feedback
system like that shown in Fig. 4c can be sufficient. However, these
specifications add constraints on exergame design that cannot be
neglected.
The VE is also the phase in which input and output require-
ments should be analyzed to choose the most adequate devices.
Very often exergames are developed around a single device, like
Wii Fit with the Wii balance board or TyroS6 with the Tymo balance
board. In a therapeutic setting, the choice of the device should come
only after the structural definition of the exercises and a rehabilita-
tion plan may require the addition of other devices to meet sec-
ondary goal requirements.
Our approach has the important byproduct of allowing the def-
inition of quantitative assessment measurements, required to eval-
uate the effect of the therapy. Usually, the outcome of exergames is
evaluated through standardized tests that require measurements
to be taken regardless of the games [39,40]; from the definition
of the goals of the exercises, quantitative measurements can be
instead computed from tracked input data and provided to clini-
cians to obtain a day-by-day quantitative evaluation of patients’
performance. This would be a large step forward in closing the loop
between exergaming and the clinicians, especially when exergam-
ing has to be performed autonomously.
Only as a last step the exergame is designed and realized. Our
methodology gives clear constraints to the game developers, who
are instructed on the areas that they are allowed to intervene,
where they can develop all the elements that improve the attrac-
tiveness of the final exergames to maximize patient compliance.
The priority should be given to allowing the patient to enter a state
of flow, defined as a state of heightened focus in which time seems
to fly by [41]. To achieve the state of flow, two aspects are consid-
ered equally important by the Dual Flow model recently proposed
in [42]. The first component requires a game to have a good bal-
ance between fitness requirements (the body ‘‘skill” in tolerating
an exercise), and exercise intensity (the challenge of the exercise
on the body). These two aspects can be regulated acting on the
parameters that are associated to the primary goal of the exercise.
However, it should be remarked that an increase in the degree of
challenge should never go against violating secondary goals of
the exercises, to prevent the exercise from becoming harmful.
The second component is related to pleasure. Nice graphics, clear
feedback, and responsiveness all contribute to this. However, we
remark that these additional aspects should be developed without
touching the primary and secondary goals of the exercise. For
example, the goal to reach state of flow should not include uncon-
strained variations in the exercise parameters related to difficulty,
or the exercise risks to be undermined. Moreover, additional game
mechanics that do not interfere with primary and secondary goals
of the exercise can be added to increase entertainment. These can
be, for instance, cognitive interaction, strategic reasoning, or role-
playing mechanics [17].
Evaluation of therapeutic games should be two-folds. On one
side, assessment tests, of clinical validity, for the disease or dys-
function targeted by the primary goal of each game, should identi-
fied. These can be, for instance, tests for postural control for games
for posture rehabilitation or classical neglect tests for neglect
6
Available at http://tyromotion.com/produkte/tyros.
64 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65

Exercise 1 Exercise 2 Exercise 3 Exercise 4 Exercise 5 Exercise 6 Exercise 7

Lateral Weight Shi Lateral Steps 360 Steps Basic Exploraon Exploraon Exploraon and
Exploraon and Cognion and Memory Target
Reposioning

VE11 VE12 VE13 VE44

VE21 VE22 VE23 VE54

VE75

VE31 VE32 VE33 VE64

Game 1 Game 2 Game 3 Game 4 Game 5 Game 6

Fruit Catcher Fire Fighter Bubbles Balloon Mix the Soup Puzzle
Burster Popper

Fig. 9. The relationship between exercises and exergames. On the top, we place exercises for balance and posture rehabilitation (left) and for neglect (right). Each exercise is
mapped into one or more VEs according to the exercise requirements. We remark that the setup of one VE can be shared with other VEs for implementation purposes, as
shown by the enclosing rectangles. This maximizes reuse. Each VE can thus be implemented into many exergames.

rehabilitation. On the other side, tests for evaluating the enjoyment
and engagement of patients are equally important for patient’s
compliance. This can be assessed, for instance, through question-
naires based on the Technology Acceptance Model, widely used
to evaluate the impact of technology [25]. The outcome of such
questionnaires on the exergames of Section 3 has demonstrated
that this kind of balance between the exercise and entertainment
has been achieved for these games.
This methodology maps well into a modular approach to thera-
peutic exergames design [25]. The VE built for a given exergame,
constituted of a bare environment (a plane, few targets, and a cam-
era) and the game mechanics can be reused for other exergames.
Similarly, quantitative assessment and parameters associated to
the degree of difficulty can be shared among all exergames associ-
ated to a given domain (Tables 1 and 2) and could be shared also
with other domains. This allows us to reuse games for different
rehabilitation goals and/or domains, thus minimizing development
effort and maximizing exploitation possibilities. An example is rep-
resented by Bubbles Burster (Figs. 4c and 7a) that can be used both
for balance and neglect rehabilitation by changing the input track-
ing device and the type of movement required. On the other hand,
an exercise can be performed through different exergames. This
creates a many-to-many relationship (Fig. 9) which can easily
increase the variety of games made available in a given domain
on one side and reuse developed games in different domains on
the other side. The benefits of reuse also extend to the exergame
components that do not depend on the game, such as a monitoring
tools. These, not being associated with any game fantasy, may be
shared by the different exergames, thus further increasing flexibil-
ity. The same is true for tools that log input data, provide manual
configuration, provide scoring and reward mechanisms, or enable
dynamic difficulty adaptation.
5. Conclusion
Separating neatly the exercise from the game in a therapeutic
exergame can be beneficial and is essential to understand the
implications of both aspects. The definition and the methodology
described here can be taken as guidelines to develop games that
can be effective and safe for use in real clinical settings and there-
fore find a wider applicability into therapeutic paths. Moreover,
modularity originating from this methodology allows maximum
reuse of the game components and of the games themselves.
The methodology presented here is potentially of broad interest
for the gaming domains in which the requirements go beyond the
mere completion of game actions, i.e. beyond entertainment.
Authors’ contributions
MP contributed to the conception, analysis, and execution of the
work, and to writing the manuscript. ES contributed to the concep-
tion, design, and implementation of games for neglect, and to writ-
ing of the manuscript. RM contributed to the conception, the
analysis, and the execution of the work. PL contributed to develop-
ing the ideas. NAB supervised the work and contributed to the con-
ception and development of the ideas, to guiding the development,
and to writing the manuscript. All authors have worked on improv-
ing the original draft and have read and approved the final
manuscript.
Conflict of interest statement
Michele Pirovano, Elif Surer, Renato Mainetti, Pier Luca Lanzi
and Nunzio Alberto Borghese declare that they have no conflict
of interest.
Financial interest statement
Michele Pirovano, Elif Surer, Renato Mainetti, Pier Luca Lanzi
and Nunzio Alberto Borghese declare that no competing financial
interests exist.
Acknowledgements
This work was partially supported by the REWIRE project
(www.rewire-project.eu), funded by the European Commission
under the FP7 framework with contract 287713.
References
[1] C. Warlow, P. Sandercock, G. Hankey, et al., Stroke: Practical Management,
Blackwell Publishing, 2008.
[2] C.J. Winstein, P.S. Pohl, R. Lewthwaite, Effects of physical guidance and
knowledge of results on motor learning: support for the guidance hypothesis,
Res. Q. Exerc. Sport 65 (4) (1994) 316–323.
[3] G. Saposnik, M. Mamdani, M. Bayley, K.E. Thorpe, J. Hall, L.G. Cohen, R. Teasell,
Clinical trial protocols effectiveness of virtual reality exercises in stroke
rehabilitation (EVREST): rationale, design, and protocol of a pilot randomized
clinical trial assessing the Wii gaming system, Int. J. Stroke 5 (2010) 47–51.
 M. Pirovano et al. / Entertainment Computing 14 (2016) 55–65
[4] J.E. Deutsch, D. Robbins, J. Morrison, P. Guarrera Bowlby, Wii-based compared
to standard of care balance and mobility rehabilitation for two individuals
post-stroke, in: 2009 Virtual Rehabilitation International Conference, June
2009, pp. 117–120.
[5] L. Prosperini, D. Fortuna, C. Giannì, et al., Home based balance training using the
Wii balance board: a cross-over pilot study in multiple sclerosis, Neurorehab.
Neural Repair (2013), http://dx.doi.org/10.1177/1545968313478484.
[6] D.A. Sparks, L.M. Coughlin, D.M. Chase, Did too much Wii cause your patient’s
injury?, J Fam. Pract. 60 (2011) 404–409.
[7] E. Flores, G. Tobon, E. Cavallaro, F.I. Cavallaro, J.C. Perry, T. Keller, Improving
patient motivation in game development for motor deficit rehabilitation, in:
Proceedings of the 2008 International Conference in Advances on Computer.
Entertainment Technology – ACE ’08, ACM Press, New York, NY, USA, 2008, p. 381.
[8] Y. Oh, S. Yang, Defining exergames & exergaming, Proc. Meaningful Play (2010)
1–17.
[9] D. Olendorf, C. Jeryan, K. Boyden, Gale Encyclopedia of Medicine, Gale Research
Company, 1999.
[10] M. Zyda, From visual simulation to virtual reality to games, Computer 38 (9)
(2005) 25–32.
[11] K. Salen, E. Zimmerman, Rules of Play: Game Design Fundamentals, 2003.
[12] M. Huber, B. Rabin, C. Docan, G.C. Burdea, M. Abdel Baky, M.R. Golomb,
Feasibility of modified remotely monitored in-home gaming technology for
improving hand function in adolescents with cerebral palsy, IEEE Trans.
Inform. Technol. Biomed. 14 (2) (2010) 526–534. A Publication of the IEEE
Engineering in Medicine and Biology Society.
[13] S. Wüest, N. Borghese, M. Pirovano, R. Mainetti, R. van de Langenberg, E. de
Bruin, Usability and effects of an exergame-based balance training program,
Games Health J. 3 (2) (2014) 106–114.
[14] R. Colombo, F. Pisano, A. Mazzone, C. Delconte, S. Micera, M.C. Carrozza, P.
Dario, G. Minuco, Design strategies to improve patient motivation during
robot-aided rehabilitation, J. Neuroeng. Rehab. 4 (2007) 3.
[15] D.J. Reinkensmeyer, C.T. Pang, J.A. Nessler, C.C. Painter, Web-based
telerehabilitation for the upper extremity after stroke, IEEE Trans. Neural
Syst. Rehab. Eng. 10 (2) (2002) 102–108.
[16] R. Lopes, R. Bidarra, Adaptivity challenges in games and simulations: a survey,
IEEE Trans. CIAIG 3 (2) (2011) 85–99.
[17] J. Schell, The Art of Game Design: Book of Lenses, Elsevier, 2008.
[18] M. LeBlanc, Eight Kinds of Fun, 2000. <http://www.8kindsoffun.com/>.
[19] W. Ijsselsteijn, H.H. Nap, Y. de Kort, K. Poels, Digital game design for elderly
users, in: Proceedings of the 2007 Conference on Future Play – Future Play ’07,
2007, pp. 17.
[20] R. Koster, A Theory of Fun for Game Design, Paraglyph Press, 2004.
[21] T.W. Malone, Heuristics for designing enjoyable user interfaces: lessons from
computer games, in: Proceedings of the ACM and National Bureau of Standards
Conference on Human Factors in Computer Systems, 1982.
[22] T.W. Bickmore, R.W. Picard, Establishing and maintaining long-term human–
computer relationships, ACM Trans. Comput. Human Interact. 12 (2) (2005)
293–327.
[23] C.D. Fisher, The effects of personal control, competence, and extrinsic reward
systems on intrinsic motivation, Organiz. Behav. Human Perform. 21 (1978)
273–288.
65
[24] J. Smeddinck, K.M. Gerling, S. Tiemkeo, Visual complexity, player experience,
performance and physical exertion in motion-based games for older adults, in:
Proceedings of the 15th International ACMSIGACCESS Conference on
Computers and Accessibility (ASSETS ‘13), ACM, New York, NY, USA, 2013, p.
8. Article 25.
[25] M. Pirovano, R. Mainetti, G. Baud-Bovy, P.L. Lanzi, N.A. Borghese, IGER –
intelligent game engine for rehabilitation, Trans. Comput. Intell. Artif. Intell.
Games (2015) (in press).
[26] L. Zimmerli, C. Krewer, R. Gassert, F. Muller, R. Riener, L. Lunenburger,
Validation of a mechanism to balance exercise difficulty in robot-assisted
upper-extremity rehabilitation after stroke, J. Neuroeng. Rehab. 9 (1) (2012) 6.
[27] H. Karnath, A.D. Milner, G. Vallar, The Cognitive and Neural Bases of Spatial
Neglect, Oxford University Press, Oxford [Oxfordshire], 2002.
[28] M. Rabuffetti et al., Touch-screen system for assessing visuo-motor
exploratory skills in neuropsychological disorders of spatial cognition, Med.
Biol. Eng. Comput. 40 (6) (2002) 675–686.
[29] R. Mainetti et al., Duckneglect: video-games based neglect rehabilitation,
Technol. Health Care 21 (2) (2013) 97–111.
[30] M.J. Riddoch, G.W. Humphreys, Perceptual and action systems in unilateral
visual neglect, Neurophysiol. Neuropsychol. Aspects Spatial Neglect (1987)
151–181.
[31] M. Bonato, Neglect and extinction depend greatly on task demands: a review,
Front. Human Neurosci. 6 (2012) 195.
[32] L. Wiart et al., Unilateral neglect syndrome rehabilitation by trunk rotation
and scanning training, Arch. Phys. Med. Rehab. 78 (4) (1997) 424–429.
[33] C.W. Johnston, L. Diller, Exploratory eye movements and visual hemi-neglect, J.
Clin. Exp. Neuropsychol. 8 (1) (1986) 93–101.
[34] C. Trepagnier, Tracking gaze of patients with visuospatial neglect, Top. Stroke
Rehab. 8 (4) (2002) 79–88.
[35] SMI-Vision, SMI Gaze & EYE Tracking Systems, 2012. <http://www.
smivision.com/> (accessed 27.01.15).
[36] The FaceTrackNoIR Website: Face Tracking with Just Your Webcam, 2010.
<http://facetracknoir.sourceforge.net/home/default.htm> (accessed 27.01.15).
[37] Logitech Inc., 3D Mouse and Head Tracker Technical Reference Manual, 1992.
<http://www.vrdepot.com/manual-tracker.pdf> (accessed 27.01.15).
[38] D. Dimovska, P. Jarnfelt, S. Selvig, G.N. Yannakakis, Towards procedural level
generation for rehabilitation, in: Proceedings of the 2010 Workshop on
Procedural Content Generation in Games (PCGames ‘10), ACM, New York, NY,
USA, 2010, p. 4. Article 7.
[39] D.J. Gladstone, C.J. Danells, S.E. Black, The Fugl-Meyer assessment of motor
recovery after stroke: a critical review of its measurement properties,
Neurorehab. Neural Repair 16 (3) (2002) 232–240.
[40] P. Azouvi, P. Bartolomeo, J.M. Beis, D. Perennou, P. Pradat-Diehl, M. Rousseaux,
A battery of tests for the quantitative assessment of unilateral neglect, Restor.
Neurol. Neurosci. 24 (4) (2006) 273–285.
[41] M. Csikszentmihalyi, Flow: The Psychology of Optimal Experience, Harper and
Row, New York, 1990. ISBN 0-06-092043-2.
[42] J. Sinclair, P. Hingston, M. Masek, Exergame development using the dual flow
model, in: Proceedings of the Sixth Australasian Conference on Computing
Education, 2004.