Dodg ing Stre R Rami Al Rih

Dodging Stre
ess Witth A Pe
ersonalized Biofeedb
back Ga
ame
Rami
R
Al Rih
hawi
Texaas A&M University
Doha, Qataar
[email protected]
Beeena Ahmed
d
Texas A&M Univ ersity
Doha,
D
Qatar
beena.ahm
[email protected]
Ricardo Gu
utierrez-Osuna
Texas A&
&M Universiity
College Staation, TX, U
USA
rgutier@
@cse.tamu.eddu
A
ABSTRACT
RELAT
TED WORK
W
We present a personalized biofeedback game
g
that trains
ssubjects to relaax during gameeplay. Training
g is achieved by
b
inncreasing the game
g
difficulty
y if the subjectt’s breathing raate
ddiffers from a prescribed
p
targ
get. Personalizaation is achieved
bby adapting gam
me difficulty to
t the subject’ss skill level, th
hus
kkeeping the gam
me challenging
g over long perriods. Validatio
on
oon a small grroup of userss indicates that the game is
eeffective at traiining players to
o acquire deep
p breathing skillls
aand reducing arrousal in a subssequent stress--inducing task.
Due too their broad appeal and iinteractive natture, video
games have been exxplored to helpp regulate the impact of
stress. In early workk, Leahy et al. [2] developedd a game to
teach deep relaxatiion to patientts with irritaable bowel
syndroome, an ailmennt to which streess is a major ccontributor.
More rrecently, Reynoolds et al. [3] ddeveloped a gaame to help
playerss cope with fe
fear; to succeeed, players hadd to lower
their ppsychological aarousal, as meeasured with a heart rate
sensor.. Parnandi et aal. [4] developeed Chill-Out, a game that
trains players to relax by penalizzing high breatthing rates
me difficultyy. When com
mpared to
with increased gam
traditioonal DB and a non-adaptivve version of the game,
Chill-O
Out was found to be more efffective in transferring DB
skills and reducing arousal durinng a subsequuent stressinducinng task.
A
Author Keywo
ords
G
Games for heallth; stress; bioffeedback; wearrable sensors.
A
ACM Classification Keyword
ds
JJ.2.3 [Computter Application
ns]: Life and Medical
M
S
Sciences –Heallth; K.8.0 [Com
mputing Milieeux]: Personal
C
Computing –Geeneral –Gamess.
INTRODUCTIO
ON
C
Chronic stress is becoming the “silent killer” due to th
he
hhidden dangerrs that it posees to our heaalth. Common
nly
pprescribed stress reduction techniques
t
incllude meditatio
on,
ddeep breathing (DB), relaxattion and biofeeedback [1], wiith
D
DB being th
he easiest and
a
most co
ommonly useed.
U
Unfortunately, these treatmen
nts have high attrition
a
rates du
ue
too decreasing motivation
m
levels with time and fail to traain
ssubjects to eng
gage in relaxaation-inducing behavior in th
he
ppresence of strressors. Thus, there is a neeed for treatmen
nts
thhat help individuals learn to
o regulate theirr stress respon
nse
uunder duress while
w
keeping th
hem motivated.
(a)
(b)
Fig. 1. (a) The goal oof the game is tto move the blu
ue ball from
one sid
de of the screen
n to the other w
without hitting the smaller
obstaclles. (b) Mappingg breathing ratee into game diffiiculty.
W
We present Dodging
D
Stress, a biofeedback game wheere
ssubjects are trrained to slow
w down theirr breathing (i.e.
bbreathe deeply
y) to induce reelaxation by modulating
m
gam
me
ddifficulty. Thee game is th
hen personalized by furth
her
aadapting game difficulty baseed on the subject’s skill level to
m
maintain engag
gement. We also present resu
ults from a pillot
sstudy to test thee effectiveness of the approacch.
GAME
E DESIGN
Dodginng Stress is aan adaptation of Dodge [5]], an open
source Android gamee under GNU-G
GPL. Shown inn Fig. 1(a),
the goaal in Dodge is to steer a ball ffrom one side of the field
to the other side w
without hitting any obstacless, as many
We adapt Dodgge for game-biofeedback
times aas possible. W
purposses by introduucing a positivve feedback ccontrol law
that inncreases the ggame difficultty in proportiion to the
player’’s breathing rrate deviating from the ideeal of five
breathss per minute. Namely, giveen the player’ss breathing
rate ( ), game difficculty (
at tim
me follows a piecewiseFig. 1(b). In coontrast with [5]], however,
linear U curve; see F
the gam
me difficulty iis not tied to aan intrinsic paarameter of
the gam
me but to the pplayer’s skill level, as measuured during
gamepplay. This allow
ws the game to adapt to eaach player,
keepinng them engaaged regardlesss of their skkill levels.
Namelly, the game m
maintains an estimate of thhe player’s
P
Permission to make digital or hard copiies of part or all of this
t work for person
nal
oor classroom use is granted without feee provided that co
opies are not made or
ddistributed for profit or commercial ad
dvantage and that co
opies bear this notiice
aand the full citation on the first page. Copyrights
C
for third
d-party components of
thhis work must be ho
onored. For all otheer uses, contact the Owner/Author.
O
C
Copyright is held by
y the owner/author(ss).
C
CHI PLAY '14, Oct 19-22 2014, Torontto, ON, Canada
A
ACM 978-1-4503-3014-5/14/10.
hhttp://dx.doi.org/10.1145/2658537.2661
1310
399
activityy. Work is undderway to com
mpare the effectiveness of
Dodginng Stress agaainst traditionaal relaxation techniques
(i.e., ddeep breathingg), and investiigate the effecct of game
difficuulty (e.g. easy vvs. hard settings) on skill traansfer on a
larger subject pool. Alternative measures of stress (i.e.
salivarry cortisol) annd game effecctiveness (i.e. subjective
experieence) are aalso being ppursued at tthis time.
Personnalization mayy also be improved byy tracking
perform
mance over reepeated attemppts to better ppredict the
subjectt’s optimal leevel. This willl maintain gaame appeal
over llonger periods, thus reduccing attrition rates and
enablinng continuous improvement iin relaxation skkills.
pprobability of success ( = successful/o
overall attemptts)
oover a 45-sec window and then adjusts the number of
oobstacles as:
2
2
2
3
3
3
∨
7
7 ∧ 7 ∧ 0.05
0.05
(1
1)
w
where
10
0 is the initiial number off obstacles an
nd
0.75 is the threshold (i.e. 75% chance of
o success).
P
PILOT USER STUDY
S
W
We evaluated Dodging Stresss through a user
u
study (N=
=5
m
male participan
nts, ages 20–23
3 years) with th
he protocol [5]::
 Training (4 min): Subjectts watched a video
v
describin
ng
hing (DB), and
d then practicced DB with an
deep breath
auditory paccing signal at 6 brpm for 2 minutes.
m
 Stressor (4 min): Subjectss performed a modified Stroo
op
t (CWT) as a pre-treatmen
nt stressor.
color word test
 Treatment (8 min): Subjeccts played Dodging Stress.
 Stressor (4 min):
m
Subjects repeated CWT
T post-treatmen
nt.
Game difficulty
Respiration rate (brpm)
11
T
To evaluate thee game’s effecttiveness in teacching relaxatio
on,
w
we recorded heeart rate variab
bility (HRV) an
nd electroderm
mal
aactivity (EDA)), both of wh
hich are proveen physiologiccal
inndicators of stress
s
[6, 7]. EDA was measured
m
with a
S
Shimmer GSR
R sensor, wherreas heart and respiration raate
w
were measured with a Zephyrr BioHarness.
9
7
5
3
7
5
3
1
‐
‐1
0
50
100
150 200 25
50
Time (seconds)
300
350
4
400
450
Fiig. 3. Respiratiion and game d
difficulty for sub
bject 1
ACKNO
OWLEDGEME
ENT
This ppublication waas made possibble by NPRP grant # 5678-2--282 from thee Qatar Natioonal Researchh Fund (a
made herein
membeer of Qatar Foundation). Thee statements m
are sollely the responssibility of the aauthors.
REFER
RENCES
1. Ricchardson, K. M
M. and Rothsteinn, H. R. Effectss of
occcupational stresss management intervention prrograms: A
metta-analysis. J. O
Occ. Health Psy
sy. 13. 1 (2008). 69-93.
2. Leaahy, A. et al. Coomputerised biiofeedback gam
mes: a new
metthod for teachinng stress managgement and its use in
irrittable bowel synndrome. J. RCP
P London 32. 6 (1998).
5522-6.
Fig. 2. Pre-- and post-task HRV (root mea
an square of
successive diffeerences in a 30-ssec window) an
nd EDA (numbeer
of skin conducttance responsess above 2 S in a 30-sec window
w)
R
RESULTS
3. Reyynolds, E. Neveermind: creatinng an entertaining
bioffeedback-enhannced game expperience to trainn users in
streess managemennt. In Proc. SIG
GGRAPH '13, A
ACM,
(20013).
F
Fig. 2 shows th
he HRV and EDA
E
pre- and post-tasks
p
for all
a
pparticipants in
n the study. Overall, subjjects showed a
ddecrease in ED
DA and an incrrease in HRV between
b
the prreaand post-tasks (both measu
ures indicativee of relaxation
n),
eexcept for subjject 5. Fig. 3 shows the traj
ajectory of gam
me
ddifficulty and breathing
b
rate for one subjeect in the stud
dy;
ggame difficulty
y closely follow
ws changes in respiration ratte,
ppeaking at the 50-sec mark, when
7. Between 170
4
425 seconds, the difficulty
y level track
ks the subjectt’s
ssuccess rate while
w
3
7. User feed
dback was also
ppositive, with subjects
s
expresssing an intereest in continuin
ng
too play the gam
me.
4. Parrnandi, A. et al.. Chill-Out: Rellaxation Traininng through
Resspiratory Biofeedback in a Moobile Casual Gaame. In
Prooc. MobiCase 22013, (2013), 2552-260.
5. Doddge by Dozing Cat Software.
httpps://code.googlle.com/p/dodgee-android/
6. Maartini, F. and Baartholomew, E.. Essentials of aanatomy
andd Physiology. B
Benjamin Cumm
mings, 2010.
7. Freeedman, L., et aal. The relationsship of sweat ggland count
to eelectrodermal aactivity. Psychoophysiology 31.. 2 (1994).
1966-200.
C
CONCLUSION AND FUTURE
E WORK
O
Our work show
ws that it is feassible to use bio
ofeedback gam
mes
too acquire relaaxation skills while
w
performiing an engagin
ng
400