This is the talk I gave at the International Conference on Virtual Rehabilitation 2013, part of the following workshop: Designing an effective rehabilitation simulation (Gerard Fluet, Wendy Powell, Sergi Bermúdez i Badia, Qinyin Qiu)
After completing this workshop participants should be able to: (1) Outline the process for developing, designing and testing an effective rehabilitation simulation, (2) Describe the process of shaping human movement abilities using simulated activities, (3) Identify variables that need to be considered when designing a rehabilitation activity, (4) Evaluate an open source virtual environment or game for conversion to a rehab activity, (5) Identify the strengths and weaknesses of commercially available software platforms, (6) Describe commonly used metrics to measure simulated movement performance, (7) Describe key features of a rehab activity that can be used to evaluate changes in movement performance. Intended audience: Computer Engineers, Biomedical Engineers, Physical Therapists and Occupational Therapists with less than three years experience in the design and development of simulated rehabilitation activities.
Unlocking the Potential of the Cloud for IBM Power Systems
ICVR2013 Workshop: Designing an effective rehabilitation simulation
1. Sergi Bermúdez i Badia
Inv. Assist. Prof., Universidade da Madeira
Marie Curie Research Fellow, Madeira – ITI
sergi.bermudez@uma.pt
2. NeuroRehabLab
http://neurorehabilitation.m-iti.org
Universidade da Madeira / Madeira – ITI
Sergi Bermúdez i Badia
Mónica S. Camerião
Athanasios Vourvopoulos
Ana Lúcia dos Santos Faria
Quality of Life Technologies Center
http://www.cmu.edu/qolt
Carnegie Mellon University
Daniel P. Siewiorek
University of Pittsburgh
Department of Occupational Therapy
Scott Bleakley
Myomo Inc
http://www.myomo.com
SPECS - http://specs.upf.edu
Universitat Pompeu Fabra
Steve Kelly
Ela Lewis
Paul F.M.J. Verschure
PERCRO - http://www.percro.org
Scuola Superiore Sant'Anna
Hospital del Mar
Medicina Física i Rehabilitació
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
Antonio Frisoli
Esther Duarte Oller
3. • Our approach (not necessarily the best one)
• The process for developing, designing and testing an
effective rehabilitation simulation
– Only Stroke
– Particular case of upper limb rehabilitation
– Not going to explain or justify the use of VR
• Metrics for simulated movement performance and
monitoring
• Personalization using the metrics
• Lessons learned from different studies
• Open source / game engines for rehabilitation
activities
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
4. 1. The process for designing, developing and
testing an effective rehabilitation simulation
hypothesis
where
- Designing defining
- Developing formulating and implementing
- Testing clinical validation
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
8. 1. Treatment frequency and intensity correlate with recovery
(Kwakkel et al., 2004; Sonoda, Saitoh, Nagai, Kawakita, & Kanada,
2004).
deployment
2. Movement practice and repetition play a fundamental role in
recovery (Karni et al., 1995).
simulated task / game mechanics
3. Specificity of rehabilitation training with respect to the deficits and
required functional outcomes has an impact on recovery (Krakauer
2006).
Interface technology & personalization
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
9. - The brain has the capability to reorganize
itself in such a way that alternate brain areas
take over other functions. The best way to
stimulate this reorganization is still under
discussion, and several approaches have been
proposed.
Dobkin, Nat ClinPractNeurol 2008
- Stroke rehabilitation should focus on
maximizing cortical reorganization.
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
10.
11. - VR can provide:
-
Fully controlled environments
-
Minimally supervised intensive
training
-
Task-specific movement reiteration
-
Individualized training
-
Feedback for reward and motivation
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
12. - Mirror Neuron System: neurons that are
active both, during the execution of goaloriented movements and during the
observation of the same action performed
by others.1-3
- There is strong evidence of the existence
of a Mirror Neuron System (MNS) in the
human brain.3-5
1
Rizzolatti & Fabbri-Destro, Exp Brain Res 2010
di Pellegrino et al., Exp Brain Res 1992; 2Gallese et al., Brain 1996; 3Rizzolatti & Fabbri-Destro, Exp Brain Res 2010;
4
Iacoboni et al., Science 1999; 5Buccino et al., Eur J Neurosci 2001
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
13. - VR can provide:
-
Fully controlled environments
-
Minimally supervised intensive
training
-
Task-specific movement reiteration
-
Individualized training
-
Feedback for reward and motivation
- Moreover:
-
MN can be activated through the
1
observation of artificial agents.
-
Evidence that a first-person
perspective is the optimal frame of
2
reference during action observation.
1
The Rehabilitation Gaming System: Spheroids
Gazzola et al., Neuroimage 2007; 2Maeda et al., J Neurophysiol 2000
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
14.
15. BRAIN
- Generation of Sensory Motor Rhythms
- Use of remaining Cortico-Spinal Tracts
VR
- Close the sense-act loop
- Activation of Mirror Neuron System
- Optimal Rehabilitation Guidelines
- Feedback & motivation
Trade-off:
In general, the more specific an exercise is…
the more complex the interface is…
and the more difficult deployment is …
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
16. BRAIN
VR
- Generation of Sensory Motor Rhythms
- Use of remaining Cortico-Spinal Tracts
Active movement against gravity/friction
Body movement
Finger flexion
- Close the sense-act loop
- Activation of Mirror Neuron System
- Optimal Rehabilitation Guidelines
- Feedback & motivation
Tangible interaction
Natural User Interfaces (NUI)
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
17. BRAIN
VR
- Generation of Sensory Motor Rhythms
- Use of remaining Cortico-Spinal Tracts
Active movement against gravity/friction
- Close the sense-act loop
- Activation of Mirror Neuron System
- Optimal Rehabilitation Guidelines
- Feedback & motivation
Body movement
Finger flexion
Tangible interaction
No movement against gravity/friction
Passive assistance
Active assistance
Natural User Interfaces (NUI)
* Armeo, Hocoma
* mpower 1000, myomo inc.
Adapted / Assistive Interfaces
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
18. BRAIN
VR
- Generation of Sensory Motor Rhythms
- Use of remaining Cortico-Spinal Tracts
Active movement against gravity/friction
Body movement
Finger flexion
Tangible interaction
No movement against gravity/friction
Passive assistance
No active movement
Active assistance
Mental Imagery / Neurofeedback approaches
Natural User Interfaces (NUI)
* Armeo, Hocoma
* mpower 1000, myomo inc.
- Close the sense-act loop
- Activation of Mirror Neuron System
- Optimal Rehabilitation Guidelines
- Feedback & motivation
Interaction based on:
- Movement (body kinematics):
- Range of Movement (ROM)
- Movement speed
- Movement latency
- Movement smoothness
- Muscle synergies
- Movement Coordination
- EMG
* international 10/20 system
* Armeo, Hocoma
Adapted / Assistive Interfaces
Brain Computer Interfaces (BCI)
- Autonomic Nervous System (EDA,
respiration, HR, etc…)
- Brain activity (EEG, NIR, fMRI, etc…)
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
19. - Limiting perceived factors in elderly
population:
- Perceived barriers due to
physical limitations1,2,3
- Lack of knowledge4,5
Study:
- 8 participants with upper limb deficits
( 4 male + 4 female, M= 44.8 yo, not all
strokes)
- 3 out of 8 subjects reported previous
computer experience.
- Mouse + keyboard vs.
Key-glove comparison.
1Opalinski,
J Technology in Human Services 2001; 2Peacock
et al., Eu J of Ageing 2007; 3Ng, Educational Gerontology
2008; 4Carpenter et al., Computers in Human Behavior
2007; 5Saunders, Educational Gerontology 2004.
Rubio et al., Presence 2012
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
20. - Limiting perceived factors in elderly
population:
- Perceived barriers due to
physical limitations1,2,3
- Lack of knowledge4,5
t-test, p = .02
Study:
- 8 participants with upper limb deficits
( 4 male + 4 female, M= 44.8 yo, not all
strokes)
- 3 out of 8 subjects reported previous
computer experience.
- Mouse + keyboard vs.
Key-glove comparison.
1Opalinski,
J Technology in Human Services 2001; 2Peacock
et al., Eu J of Ageing 2007; 3Ng, Educational Gerontology
2008; 4Carpenter et al., Computers in Human Behavior
2007; 5Saunders, Educational Gerontology 2004.
Rubio et al., Presence 2012
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
21.
22. Originally developed by psychologists Robert M. Yerkes and
John Dillingham Dodson in 1908
Yerkes & Dodson , 1908
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
23. Cameirão et al., J Neuroeng Rehabil 2010
Hitting
Catching
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
Grasping
24. Automated procedure for
difficulty adjustment:
• Task individualization
• Optimized level of
performance
• Adapts to individual arms
Cameirão et al., J Neuroeng Rehabil 2010
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
25. - VR and real counterparts:
- Are they equivalent?
- Comparison with
control group
- The calibration task is used
to measure arm kinematics
and to set the baseline
parameters of the training for
every session.
- Range of Movement (ROM),
movement latency,
movement speed.
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
26. - Extracted from RGS
calibration
- Groups showed a
dissimilar pattern of
improvement in the
speed of the paretic
arm over time
(ANOVA, p<.05).
*p<.05, Mann-Whitney Test
Cameirão et al., Rest NeurNeursc, 2011.
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
27. Cameirão et al., J Neuroeng Rehabil 2010
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
28. 12 stroke patients and 10 controls were exposed to random
combinations of the 4 game parameters
4-factor ANOVA to disclose main and interaction effects
Multiple regression to estimate constants (m0 ... m9)
Cameirão et al., J Neuroeng Rehabil 2010
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
29. - Not all VR
parameters may be
relevant
- Parameters interact
in a non-trivial an nonlinear manner.
Cameirão et al., J Neuroeng Rehabil 2010
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
30. Cameirão et al., J Neuroeng Rehabil 2010
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
31. - 9 stroke patients and 10
controls
- The model captures the
behavior of the individual
arms by different game
parameters, and to adapt
the difficulty level
accordingly.
Performance
*p<.05, T-Test
Cameirão et al., J Neuroeng Rehabil 2010
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
32.
33. Training paradigm:
- Goal oriented and repetitive
actions
- Bimanual training (non-paretic
arm support)
- Parameterized (flying speed,
turning speed, acceptance radius,
distance between objects)
Bermúdez i Badia, Stroke Research & Treat, 2012
Motivation:
- Embedded in a game
- Extensive visual and sound
feedback
- Automatic computation of
training parameters (Avoid failure
and frustration)
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
34. A first study with 10 healthy participants
has shown that the NTT captures precise
quantitative kinematic information
during a NTT training session, including:
•
Range of Movement (ROM)
•
Movement smoothness
•
Arm coordination
•
Arm contribution to task
Bermúdez i Badia, Stroke Research & Treat, 2012
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
35. Kinematic measure = c0 + c1*speed + c2*turning + c3*acceptance + c4*distance
+ c5*speed*turning + c6*speed*acceptance + c7*speed*distance
+ c8*turning*acceptance + c9*turning*distance
+ c10*distance*acceptance + c11*speed2 + c12*turning2 + c13*acceptance2
+ c14*distance2
s
t
a
d
s*t
s*a
s*d
t*a
t*d
a*d
s2
t2
a2
d2
Movement
Smoothness
Range of Motion
Arm
Displacement
Arm coordination
- Not all parameters contribute to all movement kinematic measures
- We have a quantitative way of adapting parameters depending on a
desired kinematic training
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
higher level
36.
37. RGS
Control
IOT
NSG
- N = 16 acute stroke patients
- 3 weekly 20 min sessions during 12 weeks
Cameirão et al., Rest NeurNeursc, 2011.
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
38. -
RGS group recovers faster than Controls, but not at follow-up Accelerates recovery
RGS ≥ extended OT
-
Reduces therapy costs
Cameirão et al., Rest NeurNeursc, 2011.
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
39. N = 44 chronic stroke patients
5 weekly 30 min sessions during 4 weeks
1.
NUI interface
2.
Haptic interface (GRAB, PERCRO):
- Force-feedback
- Increase in the ecological validity of the task
3.
Exoskeleton (ARMEO, Hocoma):
- Support against gravity
- Control of trunk movements
Cameirão et al., Stroke, 2012.
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
40. - All groups improved
significantly
- The haptics group retained
during a longer period of
time
include if possible
- The exoskeleton group
showed a “poorer”
improvement at proximal
movements
not all assistance is
always helpful
Functional vs. correct
movement
Cameirão et al., Stroke, 2012.
Between-group
Within-group
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
*p<.05, **p<.01,
Wilcoxon Test
41. - 9 naïve healthy subjects (26.4±4.2
years)
- 3 experimental conditions
Combined motor execution and motor
imagery in VR seems to be more
effective at:
-
1
Driving cortical sensorimotor areas1,
Increasing attention and alertness to
task2,
Engaging additional related networks
(cross-modal sensory processing3 or
short term memory).
Bermúdez et al., IEEE Trans Neur Sys Reh Eng, 2013.
Solodkin et al., 2004; 2 Egner et al., 2004; 3 Kanayama et al., 2007
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
42. -18 right-handed healthy subjects
(24±3 years)
- Active and imagination conditions
using RGS
Imagery of target catching was related
to activation of frontal, parietal,
temporal, cingulate and cerebellar
regions, consistent with:
-
object processing,
motor intention,
attention,
mirror mechanisms.
Prochnow et al., Eu Journal of Neuroscience, 2013.
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
43.
44. - Virtual Environments are simulations that engage the senses of users through
real-time 3D graphics, audio and interaction to create an experience of presence
within an artificial world.
- Game Engines bring together large-scale software architectures and programming
paradigms to design and implement rendering, collision, physics and animation
processes.
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
47. Cost
~ 1500$ / developer
Platforms
All
Community
Very large &
commercial
Programming
Integrated
Development
Environment
Learning Curve
Very Fast
Customization
Using external
libraries
Unity is used in the RehabNet project
http://neurorehabilitation.m-iti.org/rehabnet
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
48. Cost
Free
~ 1500$ / developer
Free
Windows
All
Win, Mac, Linux
Large & commercial
Very large &
commercial
Developers
Programming
Development Tools
Integrated
Development
Environment
Python
Learning Curve
Fast
Very Fast
Slower
Customization
Source code
available
Using external
libraries
Source code
available
Platforms
Community
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
49. • A close dialog between health practicioners – neuroscientists –
technologists is necessary
• Our systems are not the end product, are the hypotheses
• Hypotheses need validation impact assessment
• We believe that positive effects of gaming in rehabilitation training do not
emerge out of superficial “gamification” of therapies.
• Game / training mechanics need to include:
• Neuroscientific principles of recovery
• Difficulty vs skill needs to be quantified and well understood
• Game parameters need to be automatically personalized
• Stress levels need to be controlled to ensure maximal performance
and consequent maximal learning
ICVR 2013: Designing an effective rehabilitation simulation - 26/08/2013 - Philadelphia
50. For more information contact:
sergi.bermudez@uma.pt
or visit
http://neurorehabilitation.m-iti.org