CBI MARC Review Meeting by xiuliliaofz

VIEWS: 7 PAGES: 20

									What Technologies are on the
horizon for providing care-
giving support for older adults?
Applications for furthering
independence and independent
living.

    M. Alwan et al.




        Medical Automation Research Center
What is MARC?
  MARC is a research, development and
  consulting organization providing medical and
  industrial clients with innovative automation
  solutions.
Expertise in:
  Eldercare Technologies Program: Low-cost in-
  home monitoring and assistive technologies
  Automating clinical, drug discovery, genomics,
  and health care delivery
  Collaborative multidisciplinary research and
  development
  Spinning-off small businesses
Mission of Eldercare Technologies Program
   Provide simple technological solutions:
     Enhance existing health care system, e.g.
      telehealth
     Improve quality of life for elders / disabled
     Multiply caregiver ability to interact positively
     Increase independence, mobility and levels of
      activity for older adults / the disabled
     Reduce risks and potentially reduce the costs
      of care
Smart House Project
  Adaptive, modular, low-cost,
 non-invasive monitoring
 system (suite of sensors +
 data management module)
  Service Provider Module: An
 Integrated Data Management
 System
    Remote data analysis to infer
     activities of daily living, activity
     patterns and health conditions
     over time
    Provide feedback to both
     informal and professional
     caregivers as well as health
     providers
System Overview
                                 Older adult User

   Monitoring       Collects data from
                                                    Motion
 Service Provider   multiple peripheral             Activity
                           units

       Service
                        Data
      Provider                                        Falls/Gait
                       Manager
       Server

                                                    Sleep/Bed
                                                       Exit




    Caregiver /
       Care
     Provider
Motion Sensors


Currently eight
sensors placed
in the house
detect occupant
motion
generating date
and time
stamped data.
Motion Data: Reveals Activity & Patterns
                        Sensor Firing (5:57 - 6:39 a.m.)
             8

Kitchen      7

Laundry      6
                                                                             A single day (left) can be
                                                                             examined with other days
Front Door
                                                                             and pattern activity
             5

Living Room 4
                                                                             analyzed (below)
Bathroom     3

Office       2

Bedroom      1                                              Morning Sensor Activity from One Week
             0
                 5:57     6:15                  6:21             6:39
                                    TIME




                                                                                                  8
                                                                                                 7
                                                                                                 6
                                                                                                 5
                                                                                                4
                                                       27                                       3 Sensor
                                                                                                2
                                                            25                                  1
                                                                                                0
                                                        Date      23

                                                                        21
                                                                              02
                                                                                :1
                                                                                  2   12:00 AM to 8:00 AM
Validation Results: Meal Preparation
The system was validated over 37
days, through comparisons to a
customized PDA activity log.
  No lunch or dinner events were
missed by the detection algorithm in
the validation or test data sets

   Statistic    Validation
      f          0.8600      95% Confidence
                                Interval
      k          0.8800
      p         < 0.0001
  Sensitivity    0.9063      0.7499-0.9802
  Specificity    1.0000      0.7531-1.0000
Caregiver Feedback
 Sleep Monitor                                       Matrix maps user position


The Sleep monitor is
  comprised of a suite
  of sensors that
  include:
  Matrix of Momentary
  Contact Switches
  Temperature Sensors
  Humidity and Carbon
  Dioxide Sensors
  Light Level Sensors
  Pressure Sensor

                         Processed vibration signal gives
                         pulse rate and respiration
  Validation Results:
                                                                    Heart Rate Measurements
                                                                         Vibration Sensor vs. Pulse Oximeter
                                                      70




                                    Sen sor Results
                                                      65




                                       Vibratio n

                                         (BPM)
                                                      60
                                                      55
                                                      50
                                                      45
                                                           45              50               55                60             65         70
                                                                                  Pulse Oximeter Results (BPM)
                                                                Chest, Lying on Back   Chest, Lying on Side   Chest, Lying on Stomach

                  Correlation      Standard Deviation in
  Heart Rate      Coefficient     Beats per minute (BPM)
 Measurements
                 R2       p       from best fit                 From 45° line
                                   line (BPM)                      BPM)

Chest, all      0.811   <0.0001         2.09                          2.55
positions

Abdomen, all    0.854   <0.0001         2.16                          2.23
positions

Overall         0.829   <0.0001        2.16                           2.39
Passive Unobtrusive Gait Monitor
   A highly sensitive gait
   monitor, which can be
   easily deployed in any
   home or clinical
   environment
   Small, low-cost and may
   wirelessly transmit gait
   data derived from floor
   vibrations                 The sensor system can be
                              mounted on the baseboard
   Can be used in a natural       in walkway path
   setting and does not
   require the user to do
   anything special
Preliminary Results:
Normal Gait and Fall detection
    Original Signal


    Filtered Signal



    Processed Signal
                                               Original Signal
      Timing


      Peaks


                                                 Falling Person
                                                 Detected
  Peak amplitudes: Increasing toward sensor,
              decreasing away


                                               Fall detection
Benefits of the Gait Monitor

  Measure step count and estimate average
  pace
  Distinguish between normal, limping and
  shuffling gait modes
  Detect falls
  Detect changes in pace and gait mode over
  time
  Low-cost, unobtrusive and longitudinal in-
  home gait analysis
  Potential to initiate emergency calls in case of
  falls followed by inactivity
Passive Pulse Monitor (Pipeline)
                                   Preliminary Heartbeat Detection using a Bathroom Scale
                                   using the ETC Fiber-Optic Loop-back Sensor

                            4.5

                            4.0

                            3.5

                            3.0

                            2.5
                                                                                                                     raw ETC Signal Signal
                            2.0
                            0.5


                            0.0


                            -0.5

                                                                                                   ETC signal FFT filtered from 0.8 - 1.5 Hz
                            -1.0

                            0.5

• Detects pulse rate from   0.0


barefoot person standing    -0.5

                            -1.0
on bathroom scale           -1.5
                                                                                                                     Pulse Oximeter Signal



• Blood pressure can be
                                   0     1     2     3      4     5      6      7   8    9    10    11    12    13     14   15    16     17     18     19    20

                                                                                        Time (Seconds)
                                                                                                                                 Empirical Technologies Corporation
                                                                                                                                 Charlottesville, Virginia

derived from pulse signal                                                                                                        434 296-7000




• Provides weight and
body fat percentage
Robotic Walker Project
 Laser range-finder
 Obstacle avoidance
 Front wheel steering
 control
 Passive: shared control
 platform
 Augmented with rear
 wheel collision sensor
 and audio-visual
 warnings
Benefits
 Increase independence and improve quality of life
 for the elderly
 Monitor daily activities and derive health indicators
 Opportunity for medical and / or family
 intervention before a crisis occurs
 Provide peace of mind and minimize caregiver
 burdens and strains
 Potentially delay admittance to a nursing home
 Reduce the costs of elder care
 Can be adopted by continued care facilities and
 home care providers
 Presents care providers with the opportunity to
 outreach into the community
What’s special about the MARC
monitoring concepts?
   Passive and unobtrusive
   Implements simple, low cost sensor
   technology, and computationally inexpensive
   algorithms
   Adaptive: retrofits existing home structures
   with minimal intrusion and modifications
   Employs technology that is available today
   Data mining component will yield unique
   information for the occupant, their medical
   advisors, and family members
Challenges to Adopting Technologies

 Reimbursement
 Privacy and the introduction
 of HIPAA
 Adaptability / Acceptance
 Larger scales impact studies
 (including economic and
 social ones) on elder adults,
 professional caregivers and
 informal care providers
  Thank you and Contact




Web: http://marc.med.virginia.edu
  Information request: MARCinfo@virginia.edu
  MARC Researchers:
    Robin Felder, Ph.D. rfelder@virginia.edu

    Majd Alwan, Ph.D. ma5x@virginia.edu

    Steve Kell, AAS swk3f@virginia.edu

    David Mack, BSME dcm5a@virginia.edu

    Siddharth Dalal, MSC sdalal@virginia.edu

    Beverely Turner, BS bt2h@virginia.edu

								
To top