Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1...
Transcript of Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1...
![Page 1: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/1.jpg)
1
Using Big Data to Reanimate a Paralyzed LimbSession # 187, February 22, 2017
Michael Schwemmer Ph.D., Research Statistician, Battelle
![Page 2: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/2.jpg)
2
Speaker Introduction
Michael Schwemmer, Ph.D.
Research Statistician
Battelle
![Page 3: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/3.jpg)
3
Conflict of Interest
Michael Schwemmer, Ph.D.
Salary: Salaried employee of Battelle
![Page 4: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/4.jpg)
4
Agenda
• Learning Objectives
• An Introduction of How Benefits Were Realized for the Value of Health IT
• Introduction
• Study Overview
• Neural Bypass System Components
• Demonstrations of System Use
![Page 5: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/5.jpg)
5
Learning Objectives
• Describe a system for bypassing a damaged spinal cord by using large
amounts of data collected from a cortical implant to control a muscle
stimulation system which moves a paralyzed limb controlled by the subject's
thoughts
• Examine an example of using big data to create a model that is personalized
for individuals to interpret intracortically recorded brain data and translate it
into movement
• Discuss the potential and some of the pitfalls of using big data to inform a
brain computer interface and how these lessons can be applied more
broadly
![Page 6: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/6.jpg)
6
An Introduction of How Benefits Were Realized for the Value of Health IT
• The value steps impacted were:Treatment/Clinical
Movement
Possible
Paralyzed person can
control hand movements
with thoughts
![Page 7: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/7.jpg)
7
Paralysis – Facts and Figures
• Estimated ~5.6 million people living with some form of paralysis in the United States
• Spinal cord injury (SCI) is the second leading cause of paralysis, affecting ~1.3 million individuals
• There are ~17,000 new SCI cases each year
• SCI cases alone cost the health care system ~40.5 billion annually
Causes of
paralysis
29%
23% 17%
Sources: Christopher & Dana Reeve Foundation, Paralysis and Spinal Cord Injury in the United States, 2009
National Spinal Cord Injury Statistical Center, 2016 SCI Data Sheet, 2016
![Page 8: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/8.jpg)
8
Signal Pathway From Brain to Movement
Image credit: https://en.wikipedia.org
Primary Motor Cortex
![Page 9: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/9.jpg)
9
Signal Pathway From Brain to MovementPrimary Motor Cortex
• SCI disrupts this pathway, but, in many cases, both the motor
cortex and the limb muscles that it controls remain intact
Image credit: https://en.wikipedia.org
![Page 10: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/10.jpg)
10
Bypassing the Disrupted PathwayPrimary Motor Cortex
• Idea is to bypass the
disrupted pathway and use
the intact brain signals to
control electrical stimulation of
limb musclesImage credit: https://en.wikipedia.org
![Page 11: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/11.jpg)
11
Challenges• Getting approval to perform the study
• Finding the right patient
• Constructing the muscle stimulation system
• Dealing with a massive amount of neural data
– System produces millions of data points every second
– Data processing and muscle stimulation must occur quickly in order for movements to appear natural
![Page 12: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/12.jpg)
12
Study Overview• Collaboration between Battelle and The Ohio State University
Wexner Medical Center
• FDA- and IRB-approved clinical IDE study
• Investigate the effectiveness of a cortically controlled neuromuscular stimulation to restore movement in a paralyzed person
• Study participant is a 26-yr old male who suffered a complete C5/C6 spinal cord injury from a diving accident
![Page 13: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/13.jpg)
13
Surgical Procedure
Arm Movement
Overlap
Electrode
Image courtesy: Stanford Neural Prosthetics Translational Laboratory
• Surgery performed at the Ohio State University
Wexner Medical Center ~2 years ago
• fMRI was used to map the patients’ motor
cortex prior to surgery
• A microelectrode array (MEA) was implanted in
the patients’ primary motor cortex
![Page 14: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/14.jpg)
14
System Overview
![Page 15: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/15.jpg)
15
• Neurons communicate through large changes in their electrical activity called action potentials or spikes
• The implanted microelectrode array (MEA) simultaneously records the electrical activity of neurons in a 0.15 in x 0.15 in region of primary motor cortex
Neural Data
Image credits:
https://en.wikipedia.org
https://www.codeproject.com/KB/recipes/NeuralNetwork_1/BrainNeuronSpike.png
![Page 16: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/16.jpg)
16
Neural Data
![Page 17: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/17.jpg)
17
Processing the Neural Data • The system generates 30,000 data points/sec x 96 electrodes
![Page 18: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/18.jpg)
18
Processing the Neural Data • The system generates 30,000 data points/sec x 96 electrodes
~ 2.88 million data points/sec
![Page 19: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/19.jpg)
19
Processing the Neural Data • The system generates 30,000 data points/sec x 96 electrodes (~2.88
million data points/sec)
• By isolating the signal and filtering out the noise, the data can be summarized in a much more compact form
• Big raw data can be more useful when converted to smaller more concise data
![Page 20: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/20.jpg)
20
Processing the Neural Data • The system generates 30,000 data points/sec x 96 electrodes (~2.88
million data points/sec)
• By isolating the signal and filtering out the noise, the data can be summarized in a much more compact form
• Big raw data can be more useful when converted to smaller more concise data
• Overall system requires that the data be extracted, filtered, isolated, decoded and used to trigger the muscle stimulation in 0.1 sec
![Page 21: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/21.jpg)
21
Filtering and Compressing the Data
• Filters were constructed to remove the artifact induced by the muscle stimulation
![Page 22: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/22.jpg)
22
Filtering and Compressing the Data
• Filters were constructed to remove the artifact induced by the muscle stimulation
• Wavelet decomposition was then applied to the data in order to simultaneously compress the data and construct informative features
![Page 23: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/23.jpg)
23
Wavelet Decomposition
• Wavelet decomposition has been shown to be an effective way to process raw neural signals for brain computer interface-related applications1
1Sharma et al. Time stability of multi-unit, single-unit and LFP neuronal signals in chronically implanted
brain electrodes. Bioelectronic Medicine (2015)
![Page 24: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/24.jpg)
24
Wavelet Decomposition
• Wavelet decomposition has been shown to be an effective way to process raw neural signals for brain computer interface-related applications1
• The raw signal is represented using a wavelet function basis
1Sharma et al. Time stability of multi-unit, single-unit and LFP neuronal signals in chronically implanted
brain electrodes. Bioelectronic Medicine (2015)
http://www.aticourses.com/blog/index.php/tag/continuous-wavelet-
transform/
![Page 25: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/25.jpg)
25
Wavelet Decomposition
• Wavelet decomposition has been shown to be an effective way to process raw neural signals for brain computer interface-related applications1
• The raw signal is represented using a wavelet function basis
• The coefficients of the wavelet basis functions can then be used to represent the signal
1Sharma et al. Time stability of multi-unit, single-unit and LFP neuronal signals in chronically implanted
brain electrodes. Bioelectronic Medicine (2015)
Friedenberg et al. Big Data Challenges in Decoding Cortical Activity in a Human with Quadriplegia
to inform a Brain Computer Interface. In IEEE EMBC 2016, 3084-87. Orlando, FL. (2016)
![Page 26: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/26.jpg)
26
• Data was binned into 100 msecbins (288,000 data points) and wavelet coefficients for selected scales were extracted and averaged across time and then averaged together to yield 96 data points per bin
Filtering and Compressing the Data
117 - 234 Hz
234 - 469 Hz
469 - 938 Hz
938 - 1875 Hz
![Page 27: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/27.jpg)
27
Filtering and Compressing the Data
• Data was binned into 100 msec bins (288,000 data points) and wavelet coefficients for selected scales were extracted and averaged across time and then averaged together to yield 96 data points per bin
• This feature engineering step makes it possible to process the big data in real-time with standard hardware
2.88 million data points/sec 960 data points/sec
![Page 28: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/28.jpg)
28
Decoding: Translating the Neural Data into Movement Commands
• The compressed 96-dimensional signal is
then input into our decoding (aka
classification) algorithms
• These algorithms translate brain activity
to imagined movement
• Training data is acquired by having the
subject imagine the movements of an
animated hand he is seeing on a screen
• The decoding algorithms are trained and
then used to control muscle stimulation
![Page 29: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/29.jpg)
29
Decoding: Translating the Neural Data into Movement Commands
http://www.med.nyu.edu/chibi/sites/default/files/chibi/Final.pdf
• We use a custom regularized Support
Vector Machine (SVM) decoding
algorithm (Humber et al.,2012)
![Page 30: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/30.jpg)
30
Decoding: Translating the Neural Data into Movement Commands
http://www.med.nyu.edu/chibi/sites/default/files/chibi/Final.pdf
• We use a custom regularized Support
Vector Machine (SVM) decoding
algorithm (Humber et al.,2012)
![Page 31: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/31.jpg)
31
Decoding: Translating the Neural Data into Movement Commands
• Individual decoders were trained for each
movement tested (one against all others)
• The decoder with the highest output score
above zero was used to drive the
stimulator
![Page 32: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/32.jpg)
32
Muscle Stimulation• Electrode sleeve around the participants
forearm allows for different patterns of
muscle stimulation corresponding to
different desired movements
![Page 33: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/33.jpg)
33
Overall System
![Page 34: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/34.jpg)
34
Individual Movement Task
![Page 35: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/35.jpg)
35
Individual Movement Task
• Participant was able to perform
6 different movements with an
overall accuracy of ~70%
Bouton et al. Restoring Cortical Control of Functional
Movement in a Human with Quadriplegia. Nature 533-
7602:247–250 (2016)
![Page 36: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/36.jpg)
36
Functional Improvements
![Page 37: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/37.jpg)
37
Functional Improvements
The participant was able to
successfully complete the
grasp-pour-and-stir task 3 out of
5 times in 10 minutes with a
completion time of 42 ± 10 s
![Page 38: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/38.jpg)
38
GRASSP Clinical Assessment
• The GRASSP1 is a standardized test with excellent inter-rater and test-retest reliability developed to assess sensorimotor impairment of cervical SCI patients
(Graded and Redefined Assessment of Strength, Sensibility, and Prehension)
1Kalsi-Ryan et al. Development of the Graded Redefined Assessment of
Strength, Sensibility and Prehension (GRASSP): reviewing
measurement specific to the upper limb in tetraplegia. Journal of
neurosurgery. Spine 17, 65-76 (2012)
Bouton et al. Restoring Cortical Control of Functional
Movement in a Human with Quadriplegia. Nature 533-
7602:247–250 (2016)
![Page 39: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/39.jpg)
39
GRASSP Clinical Assessment
• The GRASSP1 is a standardized test with excellent inter-rater and test-retest reliability developed to assess sensorimotor impairment of cervical SCI patients
• Motor impairment improved from the fifth and the sixth cervical (C5–C6) to the seventh cervical to first thoracic (C7–T1) level unilaterally, conferring on him the critical abilities to grasp, manipulate, and release objects
(Graded and Redefined Assessment of Strength, Sensibility, and Prehension)
1Kalsi-Ryan et al. Development of the Graded Redefined Assessment of
Strength, Sensibility and Prehension (GRASSP): reviewing
measurement specific to the upper limb in tetraplegia. Journal of
neurosurgery. Spine 17, 65-76 (2012)
Bouton et al. Restoring Cortical Control of Functional
Movement in a Human with Quadriplegia. Nature 533-
7602:247–250 (2016)
![Page 40: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/40.jpg)
40
GRASSP Clinical Assessment
• The GRASSP1 is a standardized test with excellent inter-rater and test-retest reliability developed to assess sensorimotor impairment of cervical SCI patients
• Motor impairment improved from the fifth and the sixth cervical (C5–C6) to the seventh cervical to first thoracic (C7–T1) level unilaterally, conferring on him the critical abilities to grasp, manipulate, and release objects
• These findings quantify the reduction in functional motor impairments possible with the system for patients with cervical SCI
(Graded and Redefined Assessment of Strength, Sensibility, and Prehension)
1Kalsi-Ryan et al. Development of the Graded Redefined Assessment of
Strength, Sensibility and Prehension (GRASSP): reviewing
measurement specific to the upper limb in tetraplegia. Journal of
neurosurgery. Spine 17, 65-76 (2012)
Bouton et al. Restoring Cortical Control of Functional
Movement in a Human with Quadriplegia. Nature 533-
7602:247–250 (2016)
![Page 41: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/41.jpg)
41
Credit Card Task
![Page 42: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/42.jpg)
42
Keeping the Participant Engaged
![Page 43: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/43.jpg)
43
Keeping the Participant Engaged
![Page 44: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/44.jpg)
44
Summary and Additional Challenges• Discussed the construction of our neural bypass system and the big
data challenges we encountered
• Demonstrated that SCI can be bypassed in order to allow a paralyzed individual to regain hand and wrist usage
• Neural signals are nonstationary which force the decoding algorithms to be retrained every session in order to maintain similar performance levels
• Stimulation sleeve requires recalibration
• The system has a lot of large components which currently prohibits home usage
![Page 45: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/45.jpg)
45
Future Directions• Developing robust decoding algorithms that do not require daily
retraining
– Deep learning methods
• Developing a new stimulation sleeve that easier to put on/take off, contains more electrodes, and will also keep track of the users’ hand position
• Shrinking the size of various components of the system
– Stimulator
• Recruiting the next patient and performing a new study
– Leverage and build upon the framework developed with the first patient
![Page 46: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/46.jpg)
46
Benefits Realized for the Value of Health IT
• The value steps impacted were:Treatment/Clinical
Movement
Possible
Paralyzed person can
control hand movements
with thoughts
![Page 47: Using Big Data to Reanimate a Paralyzed Limb · brain computer interface-related applications1 •The raw signal is represented using a wavelet function basis •The coefficients](https://reader034.fdocuments.in/reader034/viewer/2022052005/6018565a33bed50da65ce025/html5/thumbnails/47.jpg)
47
AcknowledgementsBattelle
– David Friedenberg
– Nicholas Annetta
– Gaurav Sharma
– Stephanie Kute
– Nick Skomrock
– Vimal Buck
– Fritz Eubanks
– Jeff Friend
– Brad Glenn
– Mingming Zhang
– Sam Colachis
– Andrew Landgraf
– Herb Bresler
– Chad Bouton (currently at Feinstein Institute)
OSU Wexner Medical Center
– Ali Rezai
– Jerry Mysiw
– Marcie Bockbrader
– Per Sederberg