Ubiquitous Health:
Wearable Computing Systems that Promote Healthy Living and Transform Health Care
The fast-growing costs of acute care are pushing the healthcare systems worldwide to a limit. Globally, we are coming to realize that we cannot afford to provide everybody with access to unlimited healthcare services in the light of current demographic changes. An alternative approach is emerging that focuses on “keeping people healthy” through primary and secondary prevention in all phases of life. This paradigm shift in the healthcare systems is demanding research in ambient, sensor-enhanced assistive technologies that “keep people outside of the hospital”. Therefore, a fast-growing interest exists for wearable and pervasive computing systems and ambient assistive technology that aim at ubiquitous health promotion for individuals in the home and community settings.
The talk will present several examples for associated research projects in the fields of sports, health, and medicine. A particular example is the miLife research project (Fig. 1). In this project, we i) implemented ambient sensors for physiological (ECG, EMG, ...) and biomechanical (accelerometer, gyroscope, ...) data recording, ii) used pervasive computing systems (e.g. in smartphones or smarthomes) for monitoring and signal processing, and iii) employed data base technology, machine learning algorithms, and simulation models in order to provide accurate information to sportsmen, patients, and caregivers in numerous applications that aimed at promoting healthy living and improving health care.
The talk will also present further research challenges that exist in the field of wearable and pervasive computing systems for ubiquitous health support. Example challenges are the required signal processing and machine learning algorithms that need to be computationally efficient yet sufficiently accurate, but also comprehensive databases, simulative data analysis and holistic data mining strategies. The outlook of the presentation will focus on future research directions that aim at contributing to the above mentioned paradigm shift in global healthcare systems by the use of wearable and pervasive computing systems for ubiquitous health support.
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Bjoern Eskofier: Keynote at DSAI & TISHW 2016 Conference
1. Ubiquitous Health:
Wearable Computing Systems that Promote
Healthy Living and Transform Health Care
Prof. Bjoern Eskofier, PhD
Endowed Professorship of the adidas AG
Digital Sports & Health Lab
December 1, 2016
5. Digital Sports Group
Digital Sports Group
Data Mining
Biomechanics
Physiology
Wearable Systems
Sensors
Algorithms
5
Sports
Applications
Biomedical
Applications
“… to increase
human health …”
8. Origins – adidas_1 (2008)
8
Eskofier et al.: Embedded Surface Classification in Digital Sports. Pat Rec Let 30(16), 2009
9. Internet
Human-Machine-Interface
(Speech, Display, Vibration,…)
M.D. AthleteCoach
Apps for Live-Feedback,
Updates miFitness
miTeam
Web 2.0
miHealth
miCoachBluetooth
ZigBee
ANT
ANDROID Mobile
Sensor Framework ASTRUM miLife
WebService
Feedback, Monitoring
and Social NetworkingFeedback Training
Sensor
Integration
Synchronization
Communication
Volume'2'280'000'€' ' ''
European'Fund'for'Reg.'Devt.'
Follow;up'project'(2015;2018):'
“Urban'Sports”,'1'558'000'€''
miLife Research Project: 2011
9
11. Gradl, S.; Kugler, P.; Lohmüller, C.; Eskofier, B.: Real-time ECG monitoring and arrhythmia detection using Android-
based mobile devices. In: Proc. of the Int. Conf. of the IEEE EMBS (EMBC2012).
Elgendi, M.; Eskofier, B.; Dokos, S. Abbott, D.: Revisiting QRS Detection Methodologies for Portable, Wearable,
Battery-Operated, and Wireless ECG Systems. PLoS ONE 9(1), e84018, 2014.
HRV
QRS
detection
ECG signal
classifcation
HR
features
Hearty – realtime ECG analysis & arrythmia detection
Biosignal Analysis
11
12. The FitnessSHIRT
12
H Leutheuser, [...], BM Eskofier.
Textile Integrated Wearable Technologies for Sports and Medical Applications. Springer, Berlin, Germany, 2016
13. Smart shoes reach the clinic:
Wearable sensor-based instrumented gait analysis
in Parkinson’s disease
15. The patient view
‘Just imagine what
we could achieve if
we start working
together – as
equals with
different but
complementary
areas of expertise!’
15
19. Embedded Gait Analysis using Information Technology
Specific focus on Parkinson‘s Disease
Funding source
Bavarian Research Foundation
Volume 878 000 €
New funding source
FAU Emerging Fields Project
Volume 860 000 €
eGaIT Research Project: 2011
19
21. Analysis Paradigm
Movement recording
Robust Stride Segmentation Barth, [...], Klucken*, Eskofier*; EMBC 2013 & Sensors 2015
Stride Signatures
Machine Learning: Waveshape
Klucken, [...], Eskofier, Winkler; PLoS ONE 8(2), 2013
Stride Parameters
Signal Analysis: Biomechanics
Rampp, Barth [...], Klucken, Eskofier; TBME 2015
× TO
+ HS
! MS
21
22. Signal-processing-driven Stride Parameter Calculation
Rampp, Barth [...], Klucken, Eskofier; TBME 62(4), 2015
IMU Data
Accelerometer
Gyroscope
Normalization
Calibration
Invert Axes
Stride
Segmentation
msDTW
Gait Event Detection
Mid Stance (MS)
Heel Strike (HS)
Toe Off (TO)
Spatial Gait Parameters
Orientation
Estimation
(MS to MS)
Gravity
Cancellation
Zero Velocity Update
Angle
Course
De-Drifted
Integration
Distance
Estimation
Sensor
Clearance
Estimation
(SC)
Sensor-Toe-
Distance
Estimation
Stride Length
Angle Heel Strike
Angle Toe Off
Temporal Gait Parameters
Stride Time
Stance Time
Swing Time
Time HS to HS
Time HS to TO
Time TO to HS Max Toe Clearance
Toe Clearance
Estimation
Angle Dependent
Correction of SC
Stride Parameters
22
23. Timed-Up & Go Instrumentation
Angular velocity [°/s]
First TurnWalking
Second Turn Turn-to-SitWalking
Sit-to-Walk
Time [s]
24. TUG-Phases in PD patients
n = 265 PD patients, * ANOVA (0.05), post-hoc Bonferroni. Mean time (+/- SEM)
*
*
*
*
Results of the analysis
Reinfelder, S.; […]; Klucken, J.; Eskofier, B.: Timed Up-and-Go Phase Segmentation in Parkinson's Disease Patients
using Unobtrusive Inertial Sensors. EMBC 2015. 24
This$is$great,$but…$
25. C I II III C 1 2 3 C L M H
Minimum foot
clearance
Monocenter IIT
193 PD patients
145 controls
C I II III C 1 2 3 C L M H
Stride length
Gait parameter changes in PD
Schlachetzki, J.; […]; Eskofier, B.; Klucken, J.: Smart shoes reach the clinic: Wearable Sensor-based Instrumented gait
analysis in Parkinson’s disease. Lancet Neurol, under review, 2015.
H&Y UPDRS-GAIT UPDRS-III
25
26. Gait parameter changes in PD
Longitudinal measurement – intra-individual
Long term monitoring
Stride length Stance phase Swing phase
UPDRS-III Change
at follow-up visit
Schlachetzki, J.; […]; Eskofier, B.; Klucken, J.: Smart shoes reach the clinic: Wearable Sensor-based Instrumented gait
analysis in Parkinson’s disease. Lancet Neurol, under review, 2015. 26
This$is$fantas<c!$
27. Need To Go Ambulatory
Stationary lab systems Mobile sensor systems
Non-natural scenario
Limited subject numbers
Home and everyday life
Big Data!
27
Espay, A.; [...]; Klucken, J.; Eskofier, B.; [...]; Papapetropoulos, S.: Technology in Parkinson disease: Challenges and
Opportunities. Submitted to Movement Disorders 12/2015. On behalf of the MDS Taskforce on Technology.
Pasluosta, C.; Gassner, H.; Winkler, J.; Klucken, J.; Eskofier, B.: An Emerging Era in the Management of Parkinson’s
disease: Wearable Technologies and the Internet of Things. IEEE J Biomed Health Inform 19(6), 1873-1881, 2015.
28. 8 hours of unsupervised gait of PD patients
Unsupervised Gait Analysis
Single'strides'&'
Individual'raKngs'
Gait'signatures'&'
Gait'parameters'
Daytime [hour]
Reinfelder, Marxreiter, Klucken*, Eskofier*; Unpublished, in preparation for TBME
28
Time
Sync
Sensor Data
Patient Rating
29. Unsupervised Gait Analysis
ON OFF INTERMED.Motor
Fluctuations
Gait parameters
Stride length (cm)
Freezing
Gait Pattern
Daytime [hour]
29
30. Transforming Healthcare
New reimbursement paradigm:
• At present: reimbursement per prescription & treatment
• In future: reimbursement per objectively measured
treatment success?
New chronic disease management concepts:
• Present concept:
• Future concept:
6 months 6 months
variable, dep. on needs variable
30
31. Digital Biobank
Biobank of individual
signatures from a diversity
of movement disorders:
• Neurologic: Parkinson, …
• Musculoskeletal: OA, ...
Signatures consist of:
• Inertial sensor data
• Biomechanical data
• Imaging data
• Clinical scales
31
32. EU Data Platform?
Comprehensive Center for Movement Medicine
Physician / Patient
Pharma / Industry
Database
Provide Data
Controls Access
Engage Organize
32
33. EIT Health
33
Our Vision:
EIT Health is a catalyst for change.
Our community creates novel
solutions that make
healthy lives a reality for all.
Funding by EU:
2 billion / 10 years
34. EIT Health – Partners
Menno$Kok$
Interim$CLC$Director$
Belgium/Netherlands$
CLC'UK/Ireland'
CLC'France'CLC'Spain'
CLC'Belgium/Netherlands'
InnoStars'
CLC'Germany'
CLC'Scandinavia'
34