TeleMed and eHealth

1
TeleMed and eHealth06

• Lifestyle Monitoring as a Predictive Tool in
  Telecare
• Hanson, J.1, Osipovic, D.1, Hine, N.2, Amaral,
  T.2, Curry, R.3 and Barlow, J.3
• 1 The Bartlett School of Graduate Studies,
  University College London, UK
• 2 School of Computing, University of Dundee, UK
• 3 Tanaka Business School, Imperial College
  London, UK

2
What is telecare?

• Care provided remotely by means of information
  and communication technology (ICT) to people in
  their own homes. (Curry et al., 2003).
• Telecare at home can be provided by deploying a
  wide variety of sensors which monitor
• - Security and safety of the domestic environment
  
• - Personal safety
• - Vital health signs
• - Daily activities or lifestyle

Chair sensor
Door sensor
PIR
Fall detector
3
Three generations of telecare
2nd generation 1st generation
3rd generation
Prediction of possible acute situations
Automatic detection and generation of alert calls
Personal response without system intelligence

• Three generation of Telecare (Porteus and
  Brownsell 2000)
• Generations 1 and 2 are in active mode. Real time
  response to an emergency.
• Third generation is in passive mode. Continuous
  monitoring of a persons behaviour within the
  home.

4
Aims of Telecare

• The rationale behind this way of delivering
  care is an assumption that it will allow older
  people and people with longstanding health
  conditions
• to live independently in their homes for longer
• and at the same time it will save public
  resources
• 80m Preventative Technology Grant has been
  made available to local authorities in England
  for implementing telecare over the next two years.

5
Why is it needed?
Government and other official reports calling for
telecare

• Significant demographic, political and economic
  drivers ageing, longstanding illness and
  chronic conditions, care system capacity
• Wider policy agenda provides impetus focus on
  capacity, chronic disease, prevention and self
  care
• Targets. DH, Delivering C21 NHS IT Support
  (2002, reiterated by ODPM in Nov 2005) and
  Building Telecare in England (July 2005)

6
Challenges of Telecare

• Telecare often presented as an all win
  solution, but implementation faces challenges
• - technological, not just about re-engineering
  existing services but moving on from adding to
  the widely available social alarm technology to
  an intelligent sensor system
• - organisational, diverse and complex service
  involving a range of stakeholders. Need for clear
  policy and strategy
• - cost-effectiveness, requires tools for
  evaluation
• - ethical, telecare touches on issues of
  surveillance, empowerment and control. Clear
  guidance called for when prioritising need,
  offering the service, obtaining informed consent
  and activating response protocols

7
Supporting Independence

• Funding from EPSRC (EQUAL Programme)
• Interdisciplinary team of academics, charitable
  housing providers and technology manufacturers
• Overall aim is to understand the opportunities
  for and barriers to mainstreaming telecare
  services in peoples own homes
• Study conducted in two locations
• - South Yorkshire in mainstream housing
• - Devon in an extra care housing scheme (the
  focus in this presentation)
• Use of existing off the shelf technology to
  test the LSM concept.

8
History of LSM

• The concept of LSM was developed back in the late
  1990s as
• a non-intrusive, low cost technological solution
  to enhance care to older people
• reassurance for carers (formal and informal)
• proactive rather than reactive service to large
  numbers of geographically dispersed clients
• - cuts the time needed to detect potentially
  serious problems
• Widespread understanding right across the board
  that LSM is possible now, and can provide a basis
  for clinical assessment and intervention.

9
How LSM works

• LSM sensors monitor peoples habitual
  domestic movements and daily activities such as
  movement around the house.
• By constant passive monitoring of the domestic
  environment
• integrating into an intelligent LSM system,
• learning peoples routines
• recognising deviations from this norm.
• Some deviations may be interpreted as signs of
  a forthcoming crisis, in which case upon
  detecting them an alert is issued to a carer.
  Therefore the ultimate aim of LSM is to prevent a
  crisis.
• Our study exposed a number of limitations to
  this concept of LSM. I will touch upon some of
  them in this presentation.

10
Participants profiles at the start of the
research (October 2005)
11
Data and Methodological Approach

• Mixed research methodology
• - Monitoring began in January 2006 and lasted
  for 10 months.
• - Detailed floor plans prepared, taking account
  of the furniture.
• - On average 14.8 sensors were installed in each
  of the six flats.
• - Four rounds of in-depth interviews have been
  conducted.
• - All six participants received regular blood
  pressure monitoring, and two participants
  received blood sugar monitoring.
• - Half way through the project a systematic
  review of sensor output data was conducted. This
  allowed us to create a number of vignettes.
• - Vignettes represent case studies of sensor
  activity around the time of a specific, known
  event in the life of our participants, together
  with an attempt to interpret this pattern of
  activity with the benefit of available contextual
  information.

12
Busyness

• Busyness is a measure of movement within a
  dwelling, and a count of interactions with
  sensored objects. These interactions time tag
  aspects of routine and result in regular patterns
  that can be detected within periods of each day,
  within days of each week, monthly, annually and
  so forth.
• Busyness is measured by normalising the overall
  sensor activity graphs by the number of sensors
  active in the particular period of time, applying
  weightings to each sensor depending on number of
  firings.
• Busyness should not be attributed directly to
  people, but to the firing patterns of the sensors
  in their home
• Flagging up changes in the busyness of individual
  sensors or sensor arrays that indicate
  deterioration of health or well-being.

13
Example of a Vignette Emergency Hospitalisation

• Miss Evans is 84 years old and lives alone in a
  bedsit flat that she rents from the charitable
  housing provider. She has both vision and hearing
  impairment, epilepsy and a history of coronary
  heart disease. Her personality could be described
  as a bit of a worrier.
• She has a fairly regular daily routine. For
  example round about lunch time she normally takes
  an afternoon nap. As she explains herself
• I go to dinner and after dinner I have a
  little nap on the bed, not a sleep so much as to
  get my back flat because the doctor said I must
  lie flat for an hour or so during the day.

14
Sensors installed in Miss Evans flat
15
The Event Itself

• Until the day of the crisis all of Miss Evans
  blood pressure readings were within a range
  considered normal for this individual.
• On the 3rd of March 2006 quite out of the
  blue Miss Evans felt very ill, her blood
  pressure was very high and she was taken to
  hospital. She was discharged from hospital on the
  17th of March 2006.
• The hospitalisation of Miss Evans is precisely
  the type of crisis event that lifestyle
  monitoring system aims to predict and prevent. In
  this case, analysis of lifestyle monitoring
  devices was done retrospectively.

16
Bed occupancy sensor
17
(No Transcript)
18
Chair occupancy sensor
Daily firings of chair occupancy sensor
19
Hourly firings of chair occupancy sensor, 37 days
before hospitalisation
Hourly firings of chair occupancy sensor, 37 days
after hospitalisation
20
Validation of the vignette

• Although this interpretation seems to fit in
  well with the presented picture of sensor
  activity, without some form of validation it
  remains speculation.
• We therefore re-interviewed Miss Evans some
  weeks after she had returned from hospital, to
  see how she felt about the episode.
• Factors emerged which suggest that it would be
  premature to attribute the change in her activity
  pattern to a change for the better in her health
  or well being.
• - Miss Evans did not make a rapid recovery but
  rather took a long time to get better.
• - a medication change altered her routine
• - daily visits from care workers during the
  recovery period also altered the participants
  routine

21
From a retrospective to a predictive model

• The construction of vignettes is a good way of
  becoming familiar with sensor output data and
  therefore it is an appropriate technique to be
  used at the exploratory and preliminary data
  analysis stages.
• However, even when used retrospectively, the
  vignette must be used cautiously, avoiding
  over-interpretation of sensor data.
• Even it is possible to make a retrospective link
  between busyness and the build up to a crisis, we
  cannot assume that this will automatically
  translate into a predictive approach. The build
  up to the next crisis could be different.

22
Making sense of sensors

• In order to make sense of sensors one needs
  rich contextual information about the events and
  everyday activities of participants to set
  alongside the sensor data
• This is crucial both at the stage of choosing the
  right LSM sensors and when interpreting their
  output
• Obtaining such information depends on
  establishing continuous effective communication -
  a dialogue of care - which requires a rapport
  between patients and their carers.

23
Lessons for mainstream telecare

• What are the challenges if we want to move
  towards predictive telecare?
• Technological. Gathering and analysing data on
  LSM is currently time consuming and laborious. We
  need more sophisticated sensors, data mining and
  visualisation techniques. Intelligent decision
  making tools.
• Organisational. Decisions will need to be made
  about which activities are monitored, who
  monitors them and which pattern of sensor firings
  represents a norm in the case of any particular
  individual.
• Resource. Who is going to pay for the service?
  Can we measure the costs and benefits?
• Ethical. Explaining the concept, capturing user
  wishes, and obtaining informed consent.

24
LSM Lite

• There are potential benefits to LSM for both
  active / reactive and passive / predictive
  telecare.
• In active mode the signals are usually clear.
  There is either an alert or there is not.
• In passive mode the signals are both less clear
  and more open to interpretation. The real
  challenge is knowing when a weak signal indicated
  by a change in busyness is showing that something
  is going wrong.
• In the long run, LSM will probably have to
  become much more sophisticated but in the short
  term we may need to make it simpler - LSM Lite
  - selecting one or a few key sensors to monitor,
  based on individual case histories.