Content uploaded by Jeffrey Kaye
Author content
All content in this area was uploaded by Jeffrey Kaye
Content may be subject to copyright.
DEPLOYING WIDE-SCALE IN-HOME ASSESSMENT TECHNOLOGY
Jeffrey Kaye1, Tamara Hayes1, Tracy Zitzelberger1, Jon Yeargers1, Misha Pavel1, Holly
Jimison1, Nichole Larimer1, Jessica Payne-Murphy1, Eric Earl1, Kathy Wild1, Linda Boise1, Devin
Williams2, Jay Lundell3 and Eric Dishman3
1Oregon Center for Aging & Technology, Oregon Health & Science University, 2Spry Learning,
3Intel Corporation
ABSTRACT
Cognitive or motor function decline are major
causes of loss of independent living among the aged.
Several methods employing ubiquitous or unobtrusive
technologies have been proposed for application
toward in-home assessment to identify clinically
meaningful change. Most attempts at multi-dimensional
home monitoring have been on a limited scale. This
has been the result of both technical and clinical
research challenges in applying and more importantly
testing the efficacy of such methods on a community-
wide scale. We designed and implemented a system
for application to a community based clinical trial of the
efficacy of a basic sensor net (motion and contact
sensors, RF location systems, and personal home
computer interaction) to be studied in 300 homes of
independent seniors. In this manuscript we describe a
protocol to ensure several key outcomes: facilitation of
recruitment and enrollment, customized training of
elders for in-home computer use, optimized sensor net
installation, tracking of subject status and linkage to
study management software to enable on-line, real-
time testing and trouble-shooting with seniors. The
methodology suggests that large-scale unobtrusive in-
home assessment is feasible for research needed to
establish the efficacy of such systems for detection of
cognitive decline and related conditions of aging.
1. INTRODUCTION
Over 34 million world-wide are projected to be
directly affected by dementia or age-related cognitive
decline in the coming decades [1]. Improving the
outlook for these individuals and their families will
require the realization of several research milestones.
These include both improving the care and
management of people with dementia as well as
identifying disease modifying therapies and ultimately
preventions for cognitive decline. In order to improve
the ability to reach these goals methodologies for
detecting cognitive change at its earliest time-point and
tracking change over time is a critical objective.
Conventional approaches for assessment of cognitive
decline rely on clinic or office-based assessments that
are inherently time-limited and episodically conducted
between relatively long periods of time. An alternative
to this approach is to bring the evaluation into the
home setting using several monitoring technologies
thus providing opportunities for frequent or continuous
assessment. Several models and technologies have
been proposed and employed to monitor or assess a
number of health related functions in the home
environment [2]. Currently commercially available
systems are being deployed. However, few have been
formally tested in a community setting on a large-scale
over a long period of time. Their sensitivity to change is
not known. In order to begin to study this model of
cognitive and behavioral assessment we have
developed a protocol for organizing and deploying the
technology and research team necessary to conduct
this research. The goal is to scale this research up to
conducting a 36-month longitudinal study of incident
cognitive decline in up to 300 seniors living
independently in their own homes. Described in this
manuscript are the procedures to achieve this aim and
the results of a pilot study conducted to confirm the
applicability of the protocol.
2. METHODS
2.1 Outline and overview of study
The protocol was approved by the Institutional
Review Board of Oregon Health & Science University
(OHSU). Participants all provided informed written
consent. In summary, subjects were recruited from the
Portland, Oregon metropolitan area. Entry criteria
included being a man or woman age 80 or older, living
independently (living with a companion or spouse was
allowed, but not as caregiver) in a larger than one-
room (“studio”) apartment, not demented (Mini-Mental
State Examination score > 24; Clinical Dementia rating
< 0.5) and in average health for age (medical illnesses
that would limit physical participation (e.g. wheelchair
bound) or lead to untimely death over 36 months (such
as certain cancers) were exclusions). After signing
informed consent subjects were screened for inclusion
into the study. A physical and neurological examination
was completed along with a gold standard
neurocognitive assessment employing the battery of
the National Alzheimer Coordinating Center. Home
layouts were drawn and broadband Internet access
was obtained. A wireless sensor net along with study
computers were placed in the home. Subjects not living
alone were asked to wear an identifying RFID device.
Once subjects met criteria for computer literacy
(sending and receiving e-mail) they were queried
weekly using a standard set of questions with regard to
their health and activity status. Those not using a
computer were mailed the same forms. The overall
configuration of the research platform is summarized in
figure 1.
Figure 1. Diagram of the relationship of the home based
system (user PC, wireless sensors, sensor system PC and
router) to the remote distributed research operation (research
staff, activity servers and data servers).
2.2 Facilitation of recruitment and enrollment
In order to facilitate recruitment, subject enrollment
focused on seniors living in congregate housing (e.g.
continuing care retirement communities, senior
housing apartment complexes). Initial contact was
made with facility directors or managers, a formal
presentation was made to the seniors living in the
facility and a potential participant pool was created at
the end of presentations. Follow-up calls to interested
parties were made within 2 weeks of contact.
Additional subjects were identified through word of
mouth at these facilities. For the pilot phase subjects
were selected from a list of current OHSU Layton
Aging & Alzheimer’s Disease subjects who met the
pilot study criteria for age, address and past computer
use and have been followed longitudinally.
2.3. Training of elders
Subjects were instructed in use of the system and
trained according level of computer familiarity as
determined with an initial standardized assessment.
Training entailed 6 sessions over 3 weeks. Ability to
send and receive email was the criterion for computer
proficiency.
2.4. Sensor net and computer installation
To collect continuous activity data, an unobtrusive
activity assessment system was installed in the home
of each participant. Passive infrared pyroelectric
motion sensors (MS16A, x10.com) were placed in
every room at locations expected to pick up the
participant’s movements restricted to that room.
Magnetic contact sensors (DS10A, x10.com) were
placed on each door of the home to track visitors and
absences from the home. To estimate walking speed,
MS16A sensors with a restricted field of view were
installed along a hallway so they would fire only when
someone passed directly in front of them (restricted to
± 4° field of view, or about ± 6.5 cm at a distance of
90cm from the sensor). To determine who was moving
in the home, subjects wore RFID tags (Ekahau,
Finland) that used 802.11 signal strengths to determine
location. All sensors sent their data wirelessly to a
laptop computer, where the data were time-stamped
and uploaded daily to the project data center. All data
were stored in an SQL database.
2.5. On-line subject assessment and project
management
We developed a program to monitor the status of
data transfer and quality on a daily basis. The program
provides a secure web-based interface to the data,
along with data summaries and plots of activity that
can be used to detect equipment malfunction (e.g.
dead batteries), acute changes in behavior patterns
(e.g. vacations), non-compliance (e.g. failure to
complete the health status report), or failure to upload.
Each home monitoring system can also be remotely
accessed if needed for trouble-shooting or software
upgrade.
3. RESULTS
3.1 Subject characteristics, enrollment profiles
In the pilot study 36 subjects were screened and
13 subjects were enrolled from one continuing care
retirement community. Primary reasons for non-
inclusion were: disinterest, self-described health
complications that discouraged individuals from taking
on more activities, or living in a studio apartment. The
average age of subjects was 85.4 years old (age
range, 79-92 years). The ratio of men to women was
4:9. All were Caucasian. There were nine residences
monitored with four cohabitating couples and five living
alone. Of the four couples, only one had a partner that
also used the computer, two were not interested and
the other was unable due to stroke. These three were
considered ‘Motion Only’ subjects as only their motion
sensor data was collected. Of the remaining 10, six
were deemed intermediate to advanced users by the
Computer Proficiency Test, two were labeled ‘semi-
naïve’, both having completed a computer training
course within the last year, and two were ‘naïve’ to
computers. These latter four attended the six session
training course with good consistency. In addition, four
additional intermediate users attended several classes,
depending on that day’s curriculum. 2.5 months post-
enrollment, one naïve user opted to drop out of the
study due to undisclosed health conditions. The other
naïve user, 92 years old, asked that we remove her
computer due to vision complications that couldn’t be
accommodated for with additional magnifying software
or other more expensive vision aids. She continued to
be followed in the motion-only portion of the study for
several weeks, until she suffered a fall resulting in hip
fracture and moving to another facility.
3.2 System performance and monitoring
Over a 12 week period, the project team installed 9
home study systems; 2 were removed during the
course of the pilot period. In summary we deployed: 9
study laptop computers, 5 subject use computers (and
supported 3 subject owned), 9 routers, 83 motion
sensors, 10 door sensors, and 45 wireless access
points. In the course of supporting these systems there
was frequent need (21 times) to visit the homes for
adjustments. In addition there was frequent telephone
contact; research team initiated calls, 97 times; subject
initiated calls, 47 times. Sensor falls were the most
common technical problem: 7 access point sensors, 5
door sensors,1 refrigerator sensor. Two motion
sensors needed to be moved because they kept
getting bumped off the walls (suboptimal original
locations). The systems provided a total of 818 (19,632
hours) days of monitoring with 131,079 sensor firings.
The console software used for monitoring the systems
and subjects was used frequently by the study staff.
Examples of the information obtained from this
monitoring software is presented in Figures 2-3. The
computer users in the homes all completed the weekly
on-line questionnaire as to their general health and life
events status.
4. DISCUSSION
We completed a 12-week pilot study of deployment
of a scalable in-home cognitive and behavioral
assessment system. Recruitment for the pilot phase
described here was not difficult and was facilitated by
prior positive relationships with the continuing care
retirement community. Further deployments to new
communities will clearly need to pay careful attention to
building community relations and close ties to facility
management and residents. Among those enrolled the
major causes for drop out were health changes (hip
Figure 2. Screen of the Console used to track subject activity and status; Example depicting a typical
subjects monthly total activity (“X-10 events”) and mouse movements.
fracture, cognitive impairment). This is likely to be an
ongoing source of disruption in studies within this
demographic. Subjects without major life events were
very compliant. This is similar to the experience in
focused telehealth monitoring studies [3]. Subjects did
not appear to find the weekly questionnaire too
intrusive of their time. Despite the excellent response
of the enrolled subjects, certain technical challenges
remain. Many of these are predictable in the field
setting. Simple technical issues such as sensors falling
off of their placements or not initially being optimally
positioned speak to the need to build into studies a
period of test-monitoring before formal “baseline” data
is collected. It also means that staffing of research
personnel needs to be adequate to enable timely
responses to needed change. The single most
challenging barrier to rapid deployment is the
personnel time needed in set-up, computer training
and post installation adjustments. The Console
software for monitoring the project was a major aid in
facilitating these steps of the study and allowing the
study team to coordinate their efforts. Because of the
relatively high number of cohabiting couples (likely a
result of the population being healthy adults), a system
for identifying which individual’s activity data is being
acquired is necessary for this research. This adds to
the expense and effort of this research, but more
importantly with current solutions requiring a worn
sensor or device for this need, this means that subjects
face a more obtrusive and involved research regimen.
Nevertheless, among the target population it appears
that the perceived benefit of ultimately understanding
how such systems might extend or enable their
independent living outweighed any concerns about
disruptions to their daily activities. This suggests that
applying the lessons learned from this study will result
in successful deployment of this research platform for
larger longitudinal studies.
5. CITATIONS
[1] Alzheimer’s Disease International:
http://www.alz.co.uk/adi/wad/wad2004prevalence.html
(accessed February 9, 2007)
[2] Stefanov D, Bien Z, Bang W. The smart house for older
persons and persons with physical disabilities: structure,
technology arrangements, and perspectives. IEEE
Transactions on Neural Systems and Rehabilitation
Engineering, 2004. 12(2): p. 228-250.
[3] Louis AA, Turner T, Gretton M, Baksh A, Cleland JGF. A
systematic review of telemonitoring for the management
of heart failure. Eur J Heart Fail 2003; 5:583-590.
ACKNOWLEDGEMENTS: Supported by National Institute of
Health PHS grants P30-AG008017, P30-AG024978, R01-
AG024059 and Intel Corporation.
CORRESPONDENCE: Jeffrey Kaye, MD
ORCATECH: Oregon Center for Aging and Technology
Email: kaye@ohsu.edu Website: www.orcatech.org
Figure 3. Study Console view of weekly health questionnaire results used for assessing
important life events that may affect on-going monitoring; Example of a home flood.
Figure 1 asdfasdf
