This is the ground study I made for the Cliinic.me project. This is an inquiry into the history of health-themed video games.
The conclusion : wide-audience health games lack ambition. Research has shown games with a purpose can have a broad appeal for population AND bring a real benefit by teaching medical content and actually allowing for diagnosis of particular diseases to happen.
From this study, we know that a game teaching players the basis of diagnosis while being fun is possible. And trying to turn this game into a crowd-diagnosis platform is the ultimate goal to reach.
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
How to build health video games with a purpose?
1. Cliinic.meA ground study : building a fun health game with which players can
teach diagnosis to each other.
Mathieu Goudot!
!
Majeure Management et Nouvelles Technologoies!
2. Abstract: How to build health games with a purpose?
This is a tough challenge.!
!
On our research we focused our research on the potential of building a health game with a purpose. We worked on
how to build a health-themed game that would become a crowd-diagnosis platform. This project is called Cliinic.!
!
This study regroups in a short presentation all the elements that serve as a preliminary study for our gamified crowd-
diagnosis project.!
!
We start with a qualitative analysis of the Health 2.0 trend in Part I. !
!
In Part II, we discuss the definition and the limits of today’s health games, a name that gets slapped onto anything that
is even remotely good for your well-being.!
!
We then analyze two very different games, yet equally important for our research on gamified crowd-diangosis.!
!
Part III presents a research project on a proof-of-concept game to establish a crowd-diagnosis platform, for the binary
diagnosis of Malaria-infected red blood cells. The methodology and the results are going to lay the basis for the actual
research on crowd-diagnosis.!
!
Part IV is a widely popular and recent Facebook game, from the House M.D. franchise. This genuinely fun game has a
small yet present educational part to it. Its gameplay mechanics and storytelling are going to be the inspiration for our
game.!
!
Finally we will present the progress in our research and development effort on the Cliinic project, with the team of
experts it mobilized.!
3. “We can rebuild [it]. We have the technology.”1!
HEALTH 2.0
1:
The
Six
Million
Dollar
Man,
intro
scene.
4. Digital health as a value proposition
In the US, there are structural reasons for the Patient-to-Patient and ePatient
movements to rise.!
• The cost of healthcare amounts to 20% of a typical household
income1.!
• The US is the only industrialized nation with no universal healthcare
system.!
• A consultation lasts on average 7 minutes. Time before first interruption:
18 seconds2.!
• 90% of patients want to self-manage their healthcare through
technology (accessing information, refilling prescriptions, booking
appointments)3.! Figure
1:
Dave
deBronkart,
aka
epaEent
dave,
cured
his
kidney
cancer
because
he
parEcipated
in
! a
online
community
and
learned
of
a
treatment
his
doctors
didn’t
know.
He
became
a
naEonal
spokesman
to
liberate
paEent
access
to
their
own
medical
data,
thanks
to
online
technology.
The PatientsLikeMe community constitutes great research
grounds for assessing the effects of P2P healthcare:!
• 91% of patients were diagnosed before joining the site!
• 72% of patients found it helpful for better
understanding their symptoms!
• 42% found out what it was like to take a particular
treatment thanks to patients like them.!
• 37% take the decision to start a new medication
thanks to the site.!
• 12% of patients changed their physician after using
the site4.!
Figure
2:
PaEentsLikeMe,
is
a
data-‐driven
website
building
a
community
of
paEents
based
on
the
sharing
of
all
the
Digital
health
brings
a
true
value
proposi.on
to
medical
data
paEents
generate.
Filling
in
their
data,
they
can
find
similar
paEents
to
share
medicaEon
advice,and
moral
support.
online
pa.ents
and
helps
with
big
picture
problems
such
as
public
healthcare
spending.
1:
Hidden
Cost
of
Health
Care
for
Consumers
Deloi)e
2012,
2:
John
Hopkins
2004,
3:
Connected
Health
Pulse
Survey
Accenture
2012,
4:
Sharing
Health
Data
for
BeAer
Outcomes
on
PaCentsLikeMe,
Smith
and
Wicks,
Journal
of
Medical
Internet
Research
5. Digital health as a business opportunity
B2B
The key is targetting incumbents in the health sector such as big pharma and
insurance companies. At a 2011 Games for Health meeting in Palo Alto,
Epocrates and Doximity’s founder Jeff Tangney insisted on insurance
companies being the best B2B customers for digital health services,
because of their focus on innovations reducing health spending of their
customers. !
!
PatientsLikeMe for instance employed the same business model Facebook is
embracing : leveraging and monetizing user data selling it to third
parties.!
!
With a move toward pay-for-performance for US hospitals, those are
incentivized to adopt new technologies.!
! Figure
1:
B2C
! The
amount
of
funding
that
went
to
digital
health
startups
help
assess
the
opportuniEes
in
this
nascent
field.
More
than
$500M
only
for
US
health
startups
in
2011,
with
a
clear
! preference
for
B2B
models.
According to the Health 2.0 startup incubator Rock Health, the consumer
!
market for digital health amounts to 14 billion dollars, among wich 4
!
billion for health-related video games alone.!
!
!
!
!
!
!
!
!
!
!
Figure
2:
The
tradeoffs
are
the
same
in
digital
health
than
in
any
other
B2B/B2C
sector
or
industry.
Digital
Health
White
Paper
Rock
Health
2012
6. The legacy of the 1.0 digital health players
Long before the Internet 2.0, the new mobile innovations
and the social networks, there were health 1.0 startups
that focused on building Patient communities like
WebMD or on providing new services for physicians like
PracticeFusion or the later CurveDental.!
!
Building on their experience now new startups can bring
about new business models. Here is a non-extensive,
ingenuous list of new niches, made by the Rock
Health incubator, with an example of a player in each of
these niches.!
For
the
kids
For
the
old
Emers
For
the
ladies
Figure
1:
“Old
school”
programs
such
as
WeigthWatchers
found
a
second
wind
and
even
greater
success
thanks
to
the
online
Sensing
stuff
turn,
allowing
for
be)er
community
interacEons.
InnovaEons
are
not
always
at
the
core
of
all
those
companies,
but
they
all
are
digital
in
some
sense
and
hence
provide
inspiraEon
to
following
health
2.0
startups.
QuanEfy
yourself
The
new
EMR
Be)er
doc-‐paEent
relaEonship
Diagnose
or
treat
it
Mental
Health
Enhanced
care
delivery
Health
Q&A
Get
yo’
ass
in
shape
Food
&
nutriEon
In
the
cloud
Find-‐book-‐rate
a
doctor
Bill
payment,
cost
transparency
Clinical
trial
modernizaEon
Digital
Health
White
Paper
Rock
Health
2012
7. Not your Mom and Pop’s Doctor Mario.!
HEALTH IN THE VIDEO GAMES
8. How do you mix health and games?
What is a game? !
!
Games are the most interactive media available to us today. Games are designed to give
feedback to the player, to inform them on their “game state”. !
This is the first basis of fun : the feedback loop1. It’s the atom of an interactions known as
gameplay mechanic. Those gameplay mechanics then get dressed up with other elements of
fun, such as storytelling, graphical elements, multiplayer and/or competition aspects, character
customization and progression.!
The second and most important element of fun is the goal. A game is a rule-based activity that
involves challenge to reach a goal, and that gives feedback about progress towards that goal. It’s
the challenge that brings about the fun, because of the effort , it’s where the payoff is. It can be Figure
1:
A
typical
feedback
loop
in
Doom
2
:
punch
enemy
>
get
boiled down to : trying, learning, winning.! visual
reward
+
increase
score.
The
goal
of
the
game?
! Finish
it
with
the
highest
score.
“ Games can change
people’s attitude,
!
Health games tend to align game goals with health goals. Fitness games align physical
exercises with progression and in-game scoring and rewards. Therapeutic games align health
knowledge and skills with factual information needed to win the game. This is one of the first
improve their skills and benefits of health games, and why they are quite often targetted at children (see next slides).!
!
their confidence. They The problem is most game developers stop there. They remain factual and at the
can bring behavioral knowledge level. !
!
changes. ! Games are about behavioral change, not just knowledge. And this is interesting for health
Figure
2:
games. Because in most cases people know what they need to do in order to get in shape or
Quote
of
Debra
Lieberman,
PhD,
director
of
the
Health
Games
Research
naEonal
program.
manage their health. What lacks is the behavioral change. Games can instill new habits
thanks to their game mechanics. Game developers can break the barriers that refrain us from
adopting healthy behaviors.!
Can
playing
digital
games
improve
our
health?
Debra
Lieberman
TEDxAmericanRiviera
9. Health games, what are those?
The idea of linking health and video games is not new and has been time for practically as long as video games
exist. The first commercial fitness video game was Athletic World on the NES and designed to be used with
the Power Pad. The game featured olympic-styled challenges, where the player run and jumped on the pad in
the same fashion as a modern dancepad, but also had a personalization mechanic: is asked the player for the
date and the player’s name, gender and age to adapt the challenges accordingly. Since then, the
Figure
1:
commercial success of the Wii Fit, a fitness pad for the Nintendo Wii, helped install fitness video games as The
Bandai
Power
Pad
(1986),
a
beneficial for the health and well-being of player through interactive physical exertion.! precursor
of
all
fitness
games.
!
!
!
On the other hand, educational games are often focusing on conveying knowledge and education, and tend to sacrifice gameplay
mechanics to that end. Those are then closer to serious games, that we are excluding from our research scope today. Even the best
!
among those games, such as Bronkie or ReMISSION, both targeting kids, sometimes fall short of being fun, because they mainly
!
target a particular disease or condition, cutting down the universal and general aspect medicine and health have: everyone is
concerned by health problems, why not try to tackle its whole complexity?!
!
A third type of health games are too often ignored. A hospital-simulation game, Theme Hospital was completely health-themed, but with
a cartoony twist and without rigorous real-life medical content. Those few health-themed games, like Dr. Mario or the more recent
Trauma Center, made the player play in a medical environment, but without any educational part to it. The goal of the game was
pure fun. The health theme is just there to bring relevant gameplay mechanics. Those games were far better commercial successes
than educational games.!
!
Fitness
games
EducaEonal
games
Health-‐themed
games
Figure
2:
Figure
4:
Figure
3:
The
Wii
Fit
commercial
success
is
Bronkie
(1997)
was
a
game
(developed
by
Theme
hospital
(2002)
was
a
game
where
key
to
showing
fitness
video
games
Debra
Lieberman)
where
the
player
was
a
the
player
was
a
hospital
manager
and
had
is
fun
and
engaging.
A
lot
of
dinosaur
with
asthma
that
had
to
clear
the
to
cure
crazy
and
cheeky
diseases,
it’s
research
has
been
made
on
the
planet
of
dust
and
cold
flu.
ReMISSION
(2009)
become
a
classic
of
the
genre.
Trauma
subject.
had
the
player
navigate
in
3D
in
the
body
of
Center
on
the
Wii
was
a
surgeon
simulaEon
cancer
paEent
to
destroy
tumors.
game
using
the
Wii
stylus.
10. Health Games today are disinspirational
Our hypothesis is that health games today lack the reach we see video games could have.
Healthg games are either fitness games that are a fun and interactive way to adopt a healthy
behavior, or they are educational games for kids or a targeted segment of patients.!
!
But games can truly have a more important impact. Biologists tapped into a vast
computational ressource, the ability of the human brain to recognize patterns and to reason
spatially with a multiplayer online puzzle game called Foldit. It had deep gamification elements,
and turned out to be a major success, achieving computations on a much broader scale than
what would have been possible using conventional ressources1.!
!
Jane McGonigal, PhD and UC Berkelery
!
researcher, became an advocate on how gaming
can have a purpose and impact the greater
Figure
1:
good, through four positive elements : !
FoldIt
(iniEal
release
in
2008)
is
a
massively
mulEplayer
puzzle
game
harnessing
human
players
appeal
for
compeEEon
to
• blissful productivity: gamers have goals
find
the
best
stable
folding
pa)erns
for
proteins.
and objective, the feedback loop inform
them constantly on the impact they have2!
• urgent optimism gamers spend 80% of
their time failing in game worlds, yet they
stick to the challenge through resilience2!
• stronger social bonds playing a game
boosts trust among players2!
• epic meaning playing as a hero in a Figure
2:
fantasy generates awe and wonder, Jane
MCGonigal
giving
her
2010
TED
speech
on
how
gaming
positive feelings that according to research can
help
bring
the
best
in
people.
Figure
3:
SuperBeAer,
for
which
Jane
McGonigal
is
a
CreaEve
Director,
make people more likely to collaborate on a
is
a
way
to
leverage
the
posiEve
elements
of
gaming
to
reach
self-‐development
goals
for
players.
larger scale2!
We suggest that health games can and should leverage the empowering mechanics researchers have
unveiled, because of the universal appeal the health theme has. We are going to study how those
mechanics can be applied and our applied research progress in that direction.!
!
1:
PredicCng
protein
structures
with
a
mulCplayer
online
game,
Cooper,
KhaEb
et
al.
and
>
57,000
FoldIt
players,
Nature
2010,
2
:
Be
a
Gamer,
Save
the
World,
Jane
McGonigal
The
Wall
Street
Journal
2011
12. A health crowd-sourced game on Malaria
Inspired by the TapIt experiment, researchers have been driven to try to port the same kind of
experiments to the medical field, to tap crowd-sourcing’s vast resources for health.!
!
How? Combining machine learning and gamification, UCLA researchers built a telepathology
platform to see if they could approach the accuracy in diagnosis of medical professionals with
gamers that are non-medical experts, to make remote diagnostics decisions without the need of a
local medical expert. !
!
The target field of study was the analysis of microscopic analysis of biomedical specimens, and
the target disease was malaria, because of its effects on red blood cells (RBCs) and the fact
infected RBCs are recognizable with a light microscope. Malaria is still the major health problem
Figure
1:
for scores of tropical and sub-tropical countries (cause of 20% of all childhood deaths in the
The
Malaria
Diagnosis
Game
interface:
the
gamer
uses
the
region). !
syringe
to
select
infected
red
blood
cells
and
the
bucket
to
collect
the
healthy
ones.
!
!
!
!
Interestingly enough, computer vision has been tried as a means to automate malaria detection
!
in RBCs. However the extreme variations in blood smear preparations (used to photograph
!
RBCs), cell density, illumination and optical material made the task an inconclusive one. Human
vision however can abstract those variations and recognize the sane or infected patterns.!
!
The game platform was primarily meant to be the web, but was accessible through any Android-
enabled device (tablet or phone). !
!
The game validated the skill of players by providing a training phase where they were submitted
to 261 RBCs known images of which 20 were infected and identified as such in the database, but
of course not for the player. Players had to attain a 99% level of accuracy to be allowed to
Figure
2:
participate in the actual game. Then during the game roughly 20% of the RBCs were control The
hybrid
(machine+human)
diagnosis
plaoorm.
A
computer
cells, used at the end of a game as a means to score the player’s performance.! vision
algorithm
first
assesses
if
images
are
difficult
or
not
to
diagnose
(with
T
as
the
difficulty
threshold),
difficult
images
! are
fed
to
the
players,
and
then
merged
with
the
easy
ones
to
expand
the
training
database
and
reinforce
the
computer
vision
algorithm
through
machine
learning.
Distributed
Medical
Image
Analysis
and
Diagnosis
through
Crowd-‐Sourced
Games:
A
Malaria
Case
Study,
Mavandadi,
Dimitrov
et
al.,
May
2012
Public
Library
of
Science
13. The results : a successful experiment
Within the framework described above, several experiments have been conducted to assess
the efficacy of crowdsourcing over machine learning, and the hybridization of both.!
!
The crowdsourcing experiment turned out to be a true success: “The combined
accuracy of the gamer diagnoses was 99%, with sensitivity (SE) of 95.1% and specificity
(SP) of 99.4%. The positive predictive value (PPV) and negative predictive value (NPV)
were also quite high at 94.3% and 99.5% respectively”.!
!
The computer-vision algorithm without the human crowd and on the same dataset
achieved the following results : an overall accuracy of 96.3%, with SE-SP of 69.6%–99.0%,
Figure
1:
and PPV-NPV of 87.7%–96.9%.! The
effecEve
crowdsourcing
“formula”,
quoEng
the
study
:
“The
images
! are
treated
as
a
sequence
of
binary
values
that
are
broadcast
by
the
server.
The
gamers
are
effecEvely
noisy
repeaters
that
in
the
most
ideal
The hybrid experiment, with the algorithm described in figure 2 of the previous slide, case
output
the
correct
symbol
for
the
inputs
that
they
receive.
Each
allowed to increase SP and PPV by 20% and 7% respectively, but with only 10% of the repeater
transmits
its
own
noisy
version
of
the
same
input
symbol
to
a
decoder.
The
decoder
combines
all
the
received
repeater
outputs
and
RBCs sample being submitted to the crowd, the rest being analyzed by the computer vision decodes
a
final
output
zi,
which
ideally
will
be
the
correct
label/
algorithm. This helped make this game a really viable option for crowd diagnosis, reducing diagnosis
for
the
input
images.
The
repeaters
can
be
modelled
as
Binary
CommunicaEon
Channels
(top-‐ler).
pij
corresponds
to
the
probability
of
the crowd required size while increasing its efficiency. ! receiving
symbol
j
when
in
fact
symbol
i
was
transmi)ed.”
!
The last experiment increased the RBCs sample size. This experiment yielded an SE of
97.8% and an SP of 99.1%. The PPV was 96.7% and the NPV was 99.4%. This experience
boosted the PPV, which is important as the high-rate of false-positives in malaria diagnoses
is one of the major challenges in sub-Saharan Africa, according to the study.!
!
The major takeaway : the accuracy of the human crowd is within 1.25% of the
diagnosis decision made by the infectious disease expert.!
!
It’s interesting to notice this experiment was just a proof of concept, and calls for a
broadening of the method for a larger platform with more players.!
Figure
2:
! SensiEvity
is
the
most
impacted
to
the
skill
of
the
players
for
crowd-‐
For the Cliinic project, we aim at reproducing the same kind of framework for studying the diagnosis
plaoorms,
as
this
graph
shows.
Accuracy
seems
to
be
less
sensiEve.
impact of the crowd on diagnosis. However we aim at involving a greater chunk of what
makes a game a game : storytelling, fun, & gameplay progression.!
Distributed
Medical
Image
Analysis
and
Diagnosis
through
Crowd-‐Sourced
Games:
A
Malaria
Case
Study,
Mavandadi,
Dimitrov
et
al.,
May
2012
Public
Library
of
Science
14. What’s great? What’s wrong? Is it lupus? !
THE HOUSE FACEBOOK GAME:
A CASE STUDY
15. A lesson in gameplay Inner
This facebook game epitomizes one of the observation we made: solving medical loop
mysteries can be fun. House M.D. made that clear to everyone, and even before that, TV
shows like ER spurred the medical knowhow of the masses.!
!
Facebook games is a great medium for viral growth, enabling to attain a large audience Figure
1:
(we will study this in the next few slides), all the while engaging for a long duration casual Is
this
the
Malaria
mini-‐game
spin-‐off?
No,
it’s
just
the
way
player
can
make
blood
count
gamers, and incentivize repeat usage. The approach Ubisoft chose was to cradle the invesEgaEons:
a
Emed
mini-‐game.
House M.D. breath-taking storytelling inside the familiar Farmville-like isometric layout, with
building elements in your hospital, avatar customization, and less-pervasive-than-usual friend
invites. !
!
To realize diagnosis and investigations of patients, the gaming mechanics are really simple,
and amount to copies of famous mini-games or puzzle games. A cardiac arrest triggers a
mouse-bashing action (clicking many times) or a rhythm mini game (clicking in cadence).
Those mini-games are the inner feedback loop.!
! Figure
2:
The
example
of
the
pharmacy
shows
the
The progression, aside from the storytelling, is also driven by two outer feedback loops: the appointment
mechanism:
to
cure
paEents
you
need
appointment mechanism and the widening of gameplay possibilities. By restricting the to
order
pills
wEh
cash,
if
those
are
not
in
stock,
you
need
to
wait
for
the
order
to
arrive.
At
later
levels
actions with a cost of energy for every actions, and having to order medications and wait for this
can
take
up
to
a
day.
Forcing
you
to
step
back
them to arrive, this generates an appointment loop. Unlocking investigation rooms and from
your
session
and
come
back
later,
forcing
levelling up generates a sunk-cost fallacy (I invested so much time, it would be a waste not to repeat
usage
unto
players.
continue) making the player want to continue. !
!
Part of the success of the game comes from it reproducing the kind of character
dynamics found in the TV series: the player is just a mute team member in the original
cast, with House making frequent comments and piques, in a fan-service fashion. This is
pretty efficient and makes for an enjoyable gaming experience. The storytelling is really Figure
3:
good, with patient cases still inspired by Lisa Sanders (medical advisor on the show and well- The
classic
social
game
economy
and
reward
known writer of the bestseller book Every patient tells a story), creating a real will to pursue system:
energy,
experience
points,
credits
(real
currency),
cash
(in-‐game
currency),.
You
can
also
the game and go forward.! noEce
an
invesEgaEon
room
that
is
currenly
locked.
! Outer Only
by
playing
more
will
the
player
by
able
to
access
it
and
unlock
new
mini-‐games.
To conclude, this game tightly wraps efficient game mechanics together and build a health- loop
themed game with more than usual, beyond just the game setting of a hospital. But did
Ubisoft succeed in conveying the !
16. How succesful is it in conveying medical knowledge?
The game didn’t go for what was easier: they didn’t
try to “cartoonize” or “dumb-down” the medical
content of its game. The diseases are actual
diseases, and not a funny representation of them. !
!
Rather than that, it focused on making the patient
stories as compelling as it could, all the while
being rigorously exact. !
Figure
1:
! This
board
explains
the
basics
of
differenEal
diagnosis
to
players.
The first very interesting element, is by conveying
the actual methodology of modern evidence-
based medicine. The player must undertake
Figure
2:
research and investigations, to eliminate potential
As
of
the
28th
of
August
2012,
House
M.D.:
CriEcal
Cases,
the
game
had
130,000
Daily
AcEve
Users
and
over
1,200,00
diagnoses (which are established in advance for
Monthly
acEve
users.
Its
DAU/MAU
raEo
was
rather
good,
the user). This is very positive because it’s making
within
a
11
to
13%
window
according
to
this
AppData
Graph.
the inner workings of differential diagnosis to the
user.!
The second interesting element is an in-game disease database, indexing all the diseases
met during play sessions, and through which House (a figure of authority) gives some factual
information on the conditions to the player. !
!
The reaction on the forums is very positive. Some players are actually pausing game sessions
to check for the diseases they face on Wikipedia or Google. This is exactly the reaction a
health game should strive for.!
!
As far as the Cliinic project is concerned, the House M.D.:Critical Cases game constitutes an
Figure
3:
important inspiration. ! An
unheard-‐of
feature
for
a
facebook
game:
an
actual
knowledge
! base
giving
informaEon
on
all
the
in-‐game
diseases,
their
causes,
symptoms
and
treatments.
Not only are the gaming mechanics really good and polished, but also the medical
content is relevant and improves the medical knowledge, in a relative way, of its million
of players.!
18. The dev team
Wilson Mathieu Guillaume
UBC Med student,
Ariel
reviews clinical
cases and in
charge of graphic
design!
The med team
Havard CS VP of Marketing in IT graduate and CS
undergrad, lead B2C big data startup, student, lead
developer! co-founder and developer!
product manager!
Janny
UBC Med student,
drafts clinical
cases, creates
The advisors prototypes, co-
founder!
Gert Stan
UBC Language
and Literacy
Laura
Education PhD
candidate, health
literacy researcher!
UCSD CS teacher, UBC Pharmacology
crowdsourcing and teacher and Virtual
machine learning Patient expert, reviews
expert, honorary CTO! and proof-reads cases!