CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
Digital Family Portrait Field Trial:
Support for Aging in Place
Jim Rowan
Everyday Computing Lab
Georgia Institute of Technology
Atlanta, GA 30332-0280
jrowan@cc.gatech.edu
Elizabeth D. Mynatt
Everyday Computing Lab
Georgia Institute of Technology
Atlanta, GA 30332-0280
mynatt@cc.gatech.edu
ABSTRACT
these older adults as they age, given the respective decline
in working-age adults to attend to their needs and pay for
their care. The CHI community has paid particular
attention to the investigation of home technologies to
enable older adults to “age in place.” [1,8] Primarily using
ethnographic techniques, research in “aging in place” is
being carried out in the home, in assisted living facilities
and
retirement communities. [3,4,6] Ideally, novel
technologies in the home could sustain independence for
older adults wanting to stay at home while allowing
younger adults to stay in the workplace.
A growing social problem in the U.S., and elsewhere, is
enabling older adults to continue living independently, as
opposed to moving to an institutional care setting. One key
part of this complex problem is providing awareness of
senior adults’ day-to-day activities, promoting “peace of
mind” for extended family members. The Digital Family
Portrait (DFP) is one approach to providing peace of mind
that has shown promise. To date, research on the DFP has
been limited to wizard-of-oz based experiments over short
periods of time. This paper describes a DFP field trial in
which a private home was instrumented with sensors rather
than relying on input from wizard-of-oz technology. This
field trial was conducted over a period of one year between
an aging parent living alone in her own home and her adult
child living 50 miles distant.
A Wall Street Journal article this past summer described the
dilemma of working children of aging adults. Often
geographically separated due to educational and job
opportunities, working children struggle with how to care
for an aging parent from a distance [10]. Recent research
suggests that nagging concerns about the well-being of an
aging parent can prompt a worried child to move his or her
parent to an institutional care setting. This “peace of mind”
deficit stems from a lack of daily awareness that, if
available, would better inform an assessment of an older
adult’s independence and quality of life [8].
From this field trial we find that even though there was no
critical reason for the adult child to be concerned about his
mother, all involved parties found utility in the presence of
the DFP, even those family members who were not directly
involved in the field trial itself.
Author Keywords
One proposed strategy for minimizing this peace of mind
deficit is technology interventions that rely on some sort of
sensing in the home of the senior adult and a visualization
of that data for the adult child. Potential designs include the
Digital Family Portrait at Georgia Tech [8], the CareNet
from Intel Research [1], and even an Internet teapot
commercially available in Japan that tells the adult child
when her parent has made tea [5]. Despite the intuitive
appeal of these designs, many questions remain.
awareness, ubiquitous computing, light-weight interaction,
aging, visualization, home
ACM Classification Keywords
H5.m. Information interfaces and presentation (e.g., HCI):
Miscellaneous.
INTRODUCTION
Many industrialized nations are facing the prospect, and
soon the reality, of an aging population. This demographic
imbalance instigates numerous social and economic
concerns. One unresolved question is who will take care of
First, would older adults accept this form of sensing in their
homes, and more importantly, would they welcome and rely
on the sensing or would they circumvent and game the
system? External evidence points to both conclusions. A
system such as a home security system can be reassuring,
and relied on, especially for someone living alone.
However aging adults are often stereotyped as purposefully
masking any decline in abilities to avoid outside
intervention.
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise,
or republish, to post on servers or to redistribute to lists, requires prior
specific permission and/or a fee.
CHI 2005, April 2–7, 2005, Portland, Oregon, USA.
Copyright 2005 ACM 1-58113-998-5/05/0004…$5.00.
521
CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
Similar to the DFP, researchers at Intel have developed
CareNet [2], also using wizard-of-oz simulated sensing.
Intel’s CareNet addresses the needs of the elder’s support
network by displaying icons that represent the occurrence
of important events specific to the needs of the elder such
as meals, medications and outings. CareNet is intended to
support the needs of the elder’s care-giving network that
can consist of a number of persons, some who have
drastically altered their own lifestyles to provide the support
necessary to allow the progressively aging elder to age in
place. Though both the DFP and the CareNet displays to
date are wizard-of-oz driven work addressing the needs of
older adults and their families, by addressing the elders with
a greater required level of support, CareNet is intended to
be introduced in the home much later than the DFP.
Second, could these systems truly provide useful
information to the adult children. What do they need to
know? What level of complexity in the sensing is required?
The previously mentioned Japanese teapot simply reports
its use to an outside party. In contrast, some proposed
designs require relatively complex, and potentially
erroneous, activity recognition such as recognizing when
someone has prepared and consumed a meal.
Third, how would the introduction of such as system affect
the relationship between the adult child and the aging
parent? The rosy view is characterized as a sense of
increased closeness that comes from more awareness and
reduced anxiety. A not-so-rosy picture includes an adult
child, now armed with third-party information, who doesn’t
call as often to check in on things, and an aging parent who
feels spied on and neglected.
Since this previous work shows sufficient promise to
warrant taking the next step down their shared
methodological trails, our work addresses this next step, the
deployment of technology in a longer-term field trial
without the use of wizard-of-oz sensing. This field trial
involves the deployment of sensing technology in a private
home to collect sensor activity over a one-year period, a
time period that encompasses an intensive 6-week study
period and serves as the focus of this paper. This work
addresses many of the open issues described by Intel’s
CareNet research team in a recent publication:
Of course there is no one answer to these questions as each
pairing of adult child to aging parent will have different
characteristics. Some older adults will require more care
and intervention than relatively passive monitoring, and
other adults could be so fit as to resent the intrusion. A
remaining question is then, what is the sweet spot for these
types of systems, if one exists?
The purpose of this paper is to report on a single case study
that did turn out to be within this sweet spot of utility and
acceptance. Despite the quite good health of our older
adult, both parties found sufficient utility in the system to
continue using it even after we completed the year-long
research study and removed financial support for the
network connectivity. At one level, this research provides
one set of answers to the questions just posed. At another
level, it provides a path for subsequent research that aims to
explore these questions.
“An important next step is to explore what happens to the
acceptance of technologies like the CareNet Display when
sensors are introduced to fill the role of human data
collectors. Are elders comfortable living in a home filled
with sensors? Do care network members trust the data
reported by sensors? How is the network affected by sensor
or system failure? A fully working system could also enable
longitudinal deployments to uncover other unexplored
issues. What happens when the technology gets beyond any
novelty effects? How are the privacy controls used, and are
they sufficient? What social issues do technologies like the
CareNet Display introduce to the care network? Do such
technologies contribute to a reduction in communications
or visits with the elder overtime?” [1]
Previous Work
One approach to supporting peace of mind for adult
children concerned about their aging parents is the Digital
Family Portrait (DFP) developed at Georgia Institute of
Technology [7,8]. Initially based on wizard-of-oz simulated
sensing, the DFP provides a qualitative sense of an elders
daily activity (and by extension, their well-being) by
embedding an icon for each represented day into the picture
frame that surrounds the image of the elder. Like a
traditional portrait, it is designed to be hung on the wall or
propped on a mantle, blending with household decorations.
Unlike the static picture in a frame, the digital frame
changes with time, reflecting a portion of the aging parent’s
life. From general measurements of activity, to indications
of the weather, the portrait attempts to capture observations
that would naturally occur when someone lives in the same
home or next door. Unlike most awareness interfaces that
only provide a snapshot of the present and do not address
the many questions about an aging parent that derive from
trends over time, the DFP provides representations of the
past as well as the present.
The contributions in this work are three-fold. We assess the
feasibility of the Digital Family Portrait as a technology
intervention to assist family members concerned about an
older parent living alone. As an in-depth case-study, we
report on anticipated and unanticipated uses of the system.
In particular, this case study focuses on an older adult who
is quite independent and who doesn’t seem to “need”
external monitoring. However, the older adult, her adult
child, and her extended family benefit from use of the
system. Finally we describe future challenges for the
research community based on the questions that surfaced
during this field trial.
522
CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
Figure 1: Field Trial Timeline
FIELD TRIAL OVERVIEW
Field Trial Participants
For this field trial we studied two participants, an aging
parent who we will refer to as Helen and her adult son who
we will refer to as Will. This study is made up of four
major phases as indicated in Figure 1. The onset of the
study was August 2003 with the deployment of sensors in
Helen’s home. The sensing technology and layout of
sensors is described shortly. Travel by Helen and then the
arrival of winter holidays delayed the second phase of the
study until February 2004. At this time, we conducted preinterviews with Helen and Will and requested that they
each begin filling out daily diaries. Following an initial
two-week period of diary entries, we “turned on” the
Digital Family Portrait as the third phase of the study. The
daily diary during this phase asked about use of and
conversation about the DFP. After four more weeks of
daily diaries, we concluded the third phase with postinterviews. The final phase, which began concurrent with
phase 3, concluded after we had collected one year of
sensor data. As we will discuss at the end of the paper,
Helen and Will still continue to use the DFP of their own
volition and at their own expense.
The two participants in this study, Helen a 76 year old
retired postmaster and her adult son Will, live within one
hour’s drive of each other. Helen has lived in the same
home, by herself, for the past 20+ years. She reports being
in good health and still manages and controls all aspects of
her life. She drives herself, cooks, cleans, maintains her
own home (during the study she was doing touch-up
painting) and does her own yard work. Though retired
from her career as a U.S. postmaster, she still works
regularly, both inside the home as an editor for a publishing
house and outside the home as a volunteer at her church.
The pre- and post- interviews with both participants helped
us to determine the nature of the participants’ daily
routines, the participants’ sense of awareness and feelings
of connectedness one to the other, the nature and type of
communication that existed prior to the introduction of the
DFP, and the effect of the DFP on those communications.
The quotes in this paper are pulled from these interviews.
Helen keeps in touch with her family through the use of the
phone. She calls her oldest sibling, her brother in Illinois,
every day at 9pm. On those rare occasions when she cannot
reach him by phone she will call her sister, also living in
Illinois, to find out if she knows what is going on. She calls
her sister in California twice a week on either Wednesday
or Thursday and Sunday. She calls her sister in Illinois once
or twice a week outside the occasional call to check up on
her brother. She sees Will about once or twice a month
when he comes to her house to visit, do odd chores and then
go to dinner.
Aging Parent’s Social Network
While active in her church, Helen’s social network revolves
almost entirely around her family, both her siblings and her
children. She is the 3rd oldest of 4 children with a brother,
in his 80’s, living in Illinois, an older sister living in
California and a younger sister also living in Illinois. She
has 3 children that are also fairly geographically spread
with a daughter in Missouri, a daughter in Minneapolis and
her son, our participant Will, who lives on the east coast.
For Helen’s daily diary, we asked her to report her overall
level of activity for that day, whether most of that activity
had occurred in the home, whether she had been in direct
communication with Will and if they had discussed the
DFP, and her perception of how “connected” she felt to
Will. For Will’s daily diary, we asked him to report his
perception of his mother’s overall level of activity, whether
he had been in direct communication with her that day, and
his perception of his level of “awareness” of his mother as
well as his perception of how “connected” he felt to her.
Some parts of this diary worked quite well and other parts
did not as we will discuss. One helpful addition is that
Helen often annotated her diary with a short list of the
things that she did that day.
Though she has lived in the same home for 20+ years she
does not rely on her neighbors or her contacts in the
neighborhood for her social network. She professes to know
her immediate neighbors but apparently rarely has any
dealings with them outside the occasional “Hi! How are
you today?” pleasantry. She reports
“There are days that I don’t see anybody. I’m working in
here. I don’t happen to go to the store, don’t go to church
for anything.”
When asked if she would mind if neighbors might have
access to her daily activities through the DFP, she states
that as long as she has family (though they are at some
523
CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
Figure 2: Ambient display of the Digital Family Portrait
Figure 3: Detail screen of the Digital Family Portrait
distance) she can’t see why the neighbors would be
interested in it.
“I’m fairly aware… I know she’s doing proof reading and
going to church to different functions… I’m aware of when
she is traveling to go out to her sister’s in California… ”
“Well, as long as I have a family I wouldn’t especially want
anybody else to… I don’t know anybody that I am close
enough to say I would like to have that (a DFP display) or
that they would want to have it.”
He is also knows that Helen stays in touch with her siblings.
“Religiously… via the telephone… her two sisters she calls
or they call her weekly… and she religiously calls her
brother at 9 p.m. every night.”
Though she has frequent contact with her church (she does
volunteer work there on Mondays and attends on Sunday
and Wednesdays), she does not think that they would be
interested in monitoring her activities, so long as she has
family available.
Digital Family Portrait Display and Interface
The DFP persistent display (Figure 2) presents a qualitative
visualization of Helen’s daily activity that is designed to fit
naturally into a home environment alongside other family
pictures on the fireplace mantle. Leveraging a familiar
household object, the family picture in a frame, this design
leaves the photograph untouched while populating the
frame space with icons. Each icon represents a single day’s
level of activity for the 27 previous days and the current
day. The previous day’s icons provide a history of past
conditions while the current day is updated hourly to
represent the current conditions.
Will’s Perspective
After hearing a university presentation about this research,
Will approached our group about trying out the DFP. As
we learned more about his mother, namely about her good
health and independence, we were interested in
understanding Will’s motivation for using the DFP. Based
on our interviews, Will does not seem concerned on an
everyday basis of where Helen is or what she might be
doing at any minute. Neither is he very concerned about the
possibility of her hurting herself or about her not being able
to care for herself. He is concerned
The viewer can acquire more detail concerning the level of
activity for each of the days represented in the frame by
touching the icon in question. The DFP detail display
(Figure 3) presents important context information, such as
the weather conditions for that day, the sunrise and sunset
for that day, a floor plan of Helen’s home showing the
sensor locations and includes a slider for exploring the
detailed information. Located at the bottom of the detail
display is an activity graph.
“… more along the lines of touching base with her…
checking in… She used to play some golf but she’s cutting
back on it… after 18 holes she was real sore… then it got to
the point were after 9 holes she was hurting pretty bad so
her golf days are declining.”
As the child living closest to her, he keeps up with her
general activities.
The activity graph takes the 24 hour day, divides it into 96
fifteen-minute time periods, and displays the number of
sensor firings (regardless of which sensor fired) as a bar
524
CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
Figure 4: Sensor layout in Helen’s home
because they did not require any active use by Helen, they
would provide relatively clean and reliable signals, and they
would not be physically or aesthetically obtrusive in the
home. We acknowledge that difficult privacy issues come
into play with invisible sensing, for example, for visitors to
Helen’s home. However, given the low occurrence of such
visitors, our concerns about invading Helen’s home, and the
low level of information provided by the sensors, we opted
to pursue this path.
graph. The graph of the current day (in green) is displayed
over a one-hour running average of the sensor firings for
three previous similar days (in blue). In this way, the
activity graph shows what has happened that day against
the backdrop of what typically happens on that day.
By dragging the slider across the top of the activity graph,
the floor plan is animated to show sensor firings as they
occurred that day. As pictured here, Helen has come back
from shopping, and the display shows her most recent
movement from inside the garage door to the kitchen sink
as a trace from one sensor to the other.
Sensor Layout
The monitored area of Helen’s home includes a kitchen, a
dining area, a hallway, a den, a sunroom and an office area
(see Figure 4). Even though we were only interested in
collecting gross activity by totaling sensor firings, we felt it
was important to arrange the sensors logically so that the
collected data could be used in future activity
characterization. The question is, with an upper limit of 16
sensors (our sensor budget), where to best place them?
Sensing Technology
To unobtrusively collect activity data, we installed sensors
in Helen’s home that did not require her to modify her
habits in any way. We installed these sensors in the
crawlspace of the house, attached to the exposed floor
joists, in a manner that is completely invisible to anyone in
the house.
Clearly it is important to cover all exits. When there is no
movement in the house, sensor firings at exits distinguish
between a possible problem (the occupant cannot move)
and the occupant leaving the house. A second place of
general interest is the kitchen since meal preparation is
important. By placing sensors in front of the sink, the
refrigerator and the stove, we cover the classic kitchen
triangle. The sensor at the bottom of the stairs not only
indicates passage from one area to another, it indicates that
Helen has gone upstairs to her bedroom suite. Placing it
here forces this sensor to be the last to fire at the end of the
day and the first to fire at the beginning of the next day.
There is a sensor attached to the alarm system (sensor # 2),
that is set when Helen arms the alarm system. The
remaining sensors covered passage from one room to the
next. Given Helen’s good health and limitations of our
sensing technology, we did not install sensors in her
We chose a commercial sensor product called Pulsors [11]
that is built on the simple strain gauge and includes an
electronic controller that converts the strain gauges’
resistance change into the equivalent of a relay switch
closing. The strain gauge is encapsulated in a plastic case
that is easily attached to the floor joist (from the
crawlspace) using an epoxy glue. Wires from the gauges go
back to the controller. Each controller can handle four
sensors. To interface the controllers to a computer we used
a 32 port PhidgetInterfaceKit 0/16/16 board [9]. This board
allows the sensor firings coming from the controllers to be
read into a computer for identification, time-stamping and
storage through a single USB connector.
Although many sensor options are available ranging from
wearable RFID tags, to wireless motion detectors, to the
sensors we used in this study, we opted for these sensors
525
CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
For this field trial we took advantage of an existing alarm
system in the home. As it happens, the alarm system could
be used to signal our sensing infrastructure in the same way
that a sensor would trigger it. Knowing that his mom didn’t
set the alarm on her home unless she was going to be away
for a while, Will could investigate a displayed period of
absence. He could differentiate between leaving the home
for an extended period, shown by the “alarm sensor” firing
followed by an exit into the garage, and leaving the home to
do yard work, shown by the an exit into the garage to get
the lawn mower without the “alarm sensor”.
upstairs bedroom or bath. As it turned out, Will would
check for the first sensor firing at the bottom of the stairs as
a simple indication that the night had passed well.
HOW DID WILL USE THE DFP INFORMATION
One set of questions for this field trial revolve around how
the adult child (Will) used the DFP display and interface as
a representation of his mother’s activity. At the simplest
level, the DFP provides two major affordances. The first is
the ambient, persistent display of the picture framed by the
butterfly icons. The second, the interactive “detail” screen,
requires active input and supports rudimentary browsing of
the sensing data. We organize our findings based on these
affordances
“In several cases I would go look to see if she had armed
the alarm system or not because that would tell me that she
was planning to be gone for a while… as opposed to going
out to get the newspaper or cut the grass. So I could kind of
tell.”
Investigation Prompted by the Persistent Display
The DFP display in Will’s home is located in the den,
adjacent to the kitchen, a place where it is readily visible as
Will leaves for and returns from work. In interviews, Will
reported checking the display to see if anything looked out
of the ordinary as he leaves the house in the morning and as
he returns in the evening.
His mother returning from extended travel is anxiety
producing for Will. His typical plan is to repeatedly call
her house after her expected return time although he
dislikes intruding on her before she has settled back in.
Having used the activity data collected by the DFP to
determine that his mom had returned safely from her three
week trip to California, Will stated
For example, curious about the appearance of a large icon
for a particular day Will called his mom to try to determine
what had happened that day without specifically
mentioning that there was a large icon for that day.
“Yeah, when she’s traveling, I tend to worry a little more
that something could happen… when I know that there is a
risk of something travel related… I’m a little more aware of
where she ought to be at a certain time and I will try to
check in. One of the advantages of [the DFP] to me is to be
able to resolve little concerns like that before they turn into
undue anxiety over what’s going on… Your paranoia’s
going to take over if you don’t. ”
“…to see if she volunteered any change of activity
pattern… I found that she had washed the walls and was
getting ready to paint… she was moving around a lot more
and had obviously tripped the meter a few more times… the
butterfly [icon] got big.”
It is interesting to note that Will did not call his mother to
question her about a large reading on the DFP display. The
DFP information was the catalyst for the phone call, but
Will let the information come out naturally. This kind of
subtlety may be important for the acceptance of a DFP-style
monitoring system and its integration into existing social
protocols.
“… I was intending to call her about something when she
got back… rather than calling and getting her answering
machine, [once data started arriving from her home ] I had
a pretty good clue that she had been there… she actually
got back sooner than I expected so that was a kind of
pleasant thing.”
Investigations Prompted by Will’s Curiosity
AWARENESS AND CONNECTEDNESS
Another set of questions for this field trial revolve around
the dynamics of the parent and adult child relationship. We
hypothesized that two characteristics of the relationship
would change although not necessarily in a positive
direction.
First, we assumed that the adult child’s
“awareness” of his parent’s activities would increase simply
by providing more information. We also hoped that
feelings of emotional “connectedness” between both parties
would increase given this technological bridge. However,
we acknowledged that connectedness could potentially
decrease on the part of the older adult who perhaps feels
more isolated if the adult child shifts the burden of
maintaining an emotional connection too heavily on the
DFP bridge.
The detail screen of the DFP encourages different ways to
explore the sensor data. One interface failure points to
Will’s motivation for understanding his mom’s daily
activity. The detailed display for a particular day includes a
bar graph display, a floor plan of the house and a slider that,
when dragged through the 24 hour day, animates the sensor
firings on the floor plan. About using the slider Will stated
“…I was trying to get a feel for how she was moving
around the house…”
Unfortunately the volume of the data coupled with the
difficulties involved with moving the slider by hand made
this prospective use unfeasible for this field trial. Future
versions of the DFP interface will create sensor “trails” that
persist for a portion of the day and slowly fade away.
Where awareness is more based on facts (e.g. “I know Mom
goes shopping on Thursdays”), a kind of familiarity with
526
CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
compromise. Giving up some privacy in order to maintain
autonomy is a valid choice, the question is “How much
privacy must be given up in order to maintain autonomy?”
There is, however, a limit to how much and what type of
sensor technology provides the correct balance. During the
interview Helen stated
another’s schedule, connectedness is more intimate, an
emotional state of co-existence. Awareness can lead to
feelings of connectedness, though it is not necessarily a
precedent to connectedness. Photographs on a mantle are
more about feelings of connectedness than they are about
awareness. They provide emotional support as reminders of
social and familial connections, and provide a feeling of
comfort.
“Will talked at one time about having cameras up in the
corners and I wasn’t too keen on that… I just didn’t know
that I wanted a camera watching me.”
We attempted to measure changes of perceived awareness
and connectedness by asking Helen and Will to rank these
indicators using a Likert scale daily in their diary. This
measurement technique did not produce any useful results.
With Will, we simply saw a ceiling effect as his selfreported ratings of awareness and connectedness started
high before we activated the DFP interface. With Helen, it
is clear from an examination of the data that she closely tied
connectedness to whether she talked with Will on the phone
that day. Since the frequency of phone calls did not
noticeably change, there was no shift in connectedness
based on the presence of the DFP. However, interviews
with Helen are more revealing.
Our approach is to opt for sensing that does not distinguish
between occupants. It is possible to use this type of sensing
in this research because Helen lives alone in her own home
and has few outside visitors. The sensing technology has
the advantage of not requiring any active participation in
the sensing system on the part of the Helen; there were no
badges that must be worn or sensors that must be carried.
All the participant need do is to live her life as she would
normally.
The exit interview suggests that we have selected a sensor
technology that supports a reasonable balance between
privacy and autonomy.
In a manner similar to the photograph on the mantle, even
though there is no physical manifestation of the DFP in her
home, Helen reported feeling less lonely (and therefore
more connected to her family) knowing that Will was
watching out for her through the DFP.
“…I would say that I feel more comfortable knowing that
he knows that I’m moving around. He knows that there’s
something going on down here. And if he doesn’t get
something with a malfunction, he calls.”
“But, uh… if I’m feeling lonesome, I think, “Oh well, Will
knows and so then I don’t feel so lonesome.’”
Even though the sensor technology is invisible to the
participant, Helen appears to draw comfort from the
knowledge that her son is monitoring her activity.
Surprisingly, the occasional malfunction of the equipment
causes her confidence to increase in the system. Since a
malfunction cannot be distinguished from no activity, Will
made a phone call to check. This phone call acted as a
system test to Helen, demonstrating that the system is
indeed being monitored.
Her statements initially surprised us as there is no physical
reminder of the DFP system visible in her home. However
the knowledge of the system, and likely the successful
practice of having the system connect them appears to have
significantly affected Helen’s emotional response to being
alone in her home.
PRIVACY AND (TRADITIONAL) SECURITY
Additionally, the choice of sensing technology that does not
identify the person reduces the privacy concerns that arise
when a third party, not knowing that monitoring is going on
in the home, enters the monitored space since the third
party is not identified or distinguished from the participant.
Our final set of questions revolve around the cost and
benefits tradeoffs, especially for the older adults, in
assessing that the DFP system is of sufficient value to
warrant adoption. In particular we want to understand the
perceived privacy costs in comparison to the advantages
gained from knowing that someone has more information
about your general activity and well-being. We refer to this
benefit as “security,” not in the sense of information
security, but akin to the traditional use of that word, feeling
safe from harm.
Helen’s overall position about having this sensing
technology deployed in her home can be summed up by her
comment
“I don’t feel imposed upon, or spied upon or anything.”
Another indication of perceived value came from Helen
volunteering that she wished that she had the DFP when her
mother was alive.
The invisibility of our sensor technology initially arouses
concerns about privacy. After all, if Helen forgets that she
is being monitored because the monitoring technology is
invisible, then she loses a level of privacy that most people
like to maintain. We did not find this lack of a persistent
reminder to be a concern for Helen. Generally speaking in
our research, older adults are concerned about maintaining
a careful balance between privacy and autonomy. While
both are important, aging often necessitates some
“I wish it had been available when my mother was living
and I lived in all these other towns while she was back in
Illinois. (Helen’s job as a postmaster moved her from town
to town, predominantly in the mid-west). It would have been
nice because she lived alone for 25 years and went down
hill over that period of time. It would have been nice to
527
CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
Figure 5 A and B: Monday compared to four previous Mondays versus Monday compared to 4 preceding days
know that she was up and around and moving but…That’s
when I called her. The telephone got of use…”
Will’s Perspective on Level of Intrusiveness Needed
If the sensing system is acceptable to Helen, then the
subsequent question is whether the information provided by
the system is of sufficient utility for Will. Will reports
Helen and her siblings had set up an ad hoc phone call
system to keep up with their mother.
“At this stage I’m loosely connected as opposed to
intimately aware of what’s going on. I don’t think I need to
be in her life, intruding in that respect… I imagine that I
will be more aware as time goes on… but for right now I
think we are at what I consider the right level.”
“We set up a system over the last years of her life where I
called twice a week. Somebody else (one of the other of the
3 siblings) called twice a week and then somebody else
called twice a week… so we covered 6 days and we all
called on Sunday… Or at least we all were supposed to call
her on Sunday. Sometimes we did and sometimes we didn’t.
We tried to set that up so that somebody talked to her every
day. It’s kinda hard cause I was working…”
Value for the Extended Family
Although we did not interview Helen’s other children or her
siblings, we were pleasantly surprised to hear Helen report
that other family members were interested in the system:
528
CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
volunteer work on Mondays, goes to church on
Wednesdays and Sundays, and does her yard work on
Fridays. This regular pattern of activity provides the
opportunity to characterize certain days of the week, and
with this characterization, provide a baseline such that the
adult child can better determine whether or not a chosen
day is “typical” or not.
“ …(my daughter in Missouri)… I can’t remember why but
she was very interested in it (the DFP field trial). She was
very pleased that we were doing it . It puts her mind at
ease… even though it is not her (participating in the field
trial), she knows it’s in place.”
CONCLUSIONS AND FUTURE EFFORTS
The core research question for this study is: Would a family
find sufficient value in the DFP’s ability to convey the
general activity of an older adult to her adult child for them
to successfully adopt the system? The fact that Helen and
Will continue to use this system today, over a year after its
deployment, is quite encouraging. Will both used the DFP
in anticipated and unanticipated ways such as noticing
surprising activity readings, inferring when Helen was out
doing errands, and monitoring her return from travel. As
the proponent for this technological intervention, he adeptly
leveraged this new information without causing undue
concern or discomfort for his mother.
One of the simplest ways to characterize a day is to ignore
which sensors have fired and accumulate the number of
sensor firings over a particular time period. Doing this
calculation for the current day and displaying it against the
background of previous similar days of the week provides a
simple means of comparison. We took this approach in this
field trial and it appears to reasonably characterize certain
days making them distinct from other days. The best
example is to look at Mondays which has a strong pattern
(see Figure 5).
In Figure 5A, we see a single day, Monday, April 4, 2004
shown in green against the background of the 4 previous
Mondays shown in gray. In Figure 5B, we see that same
day in green but displayed against the background of the 4
previous consecutive days that include the previous
Thursday, Friday, Saturday and Sunday. The pattern of
green bars in Figure 5A clearly match the previous
consecutive Mondays better than they match the previous
consecutive days as seen in Figure 5B.
Helen’s acceptance and emotional reliance on the system is
both a little surprising and encouraging. Her volunteering
that the DFP made her feel “less lonely” and that she
wished the DFP had been available when her mother was
alive was unexpected.
Of course this study involves only one family. However,
the success of the system with an older adult as healthy as
Helen is an indication that the potential space of possible
users may be larger than initially anticipated. In our future
research, we will investigate the utility of a DFP-like
system for older adults who have more significant health
concerns.
Going back to our data, we now have the opportunity to
exhaustively examine similarities in common days (e.g. all
Mondays) versus consecutive days (e.g. Thursday, Friday,
Saturday and Sunday). Diary entries will help us determine
days that were truly unusual in comparison to typical days.
Mining Sensor Data
Trends
Although this field trial concentrates on a six-week period,
the activity data collected during the 12 months of
monitoring represents a unique and valuable set of data
worthy of further exploration. In the literature there is an
absence of research data on a person’s movement in his or
her own house that is not biased by self-report or by thirdparty observation. We are in the process of several threads
of analysis that would provide more sophisticated
capabilities for future versions of the DFP.
What appears to be the second most common concern held
by adult children about their aging parents is knowing if
there are subtle declines in capabilities or behavior.
Examples include sleeping less as well as sleeping more,
eating less, and general reductions in activity and abilities
such as climbing stairs
The data available from the DFP sensors includes the
potential for modeling the physical space (e.g. these three
sensors are part of the kitchen), however the data is
unlabeled in that there is no training data for what certain
routines “look like” in the data. The future challenge is to
determine what types of routines can be gleaned from this
data and then to identify gradual changes in those routines.
Although we do not know of any progressive declines in
Helen’s abilities during this field trial, we do know that her
behavior shifted gradually with the changes in the seasons.
We also hope to make this data available to other research
groups as a general test data set for evaluating pattern
detection algorithms.
Mondays are like Mondays
Above all, the crux of the question posed by adult children
concerned about an elderly parent is “Was today a normal
day?” Intuition tells us that certain days of the week will
have a regular pattern. People tend to pick certain days to
carry out particular tasks such as the laundry or yard work.
Add to these routines the pattern of activity that is imposed
from the outside such as work obligations, the pace of
weekly events and even the schedule of trash collection,
and life often assumes a semi-regular pace that changes
slowly with more global rhythms such as seasons.
Although retired, Helen’s schedule has patterns driven by
her preferred routines and outside commitments. She has
529
CHI 2005
׀
April 2–7 ׀Portland, Oregon, USA
PAPERS: Assistive Applications
Sensing Multiple People
activity and feeling better.
Of course this scenario
introduces more challenges for adoption and acceptability
of the system by older adults.
One potential hiccup in the data is when there are multiple
people present in the home. This scenario has the potential
to double the sensor firings in the house and to undermine
any inferences made from the data. We plan to explore the
feasibility of identifying discontinuous sensor firings as an
indication of multiple people in different rooms. If we can
detect these firings, then the next step is to work backwards
and forwards through the data to determine when the
additional person(s) entered and left the house.
Both avenues of future field trials point to the potential
utility of the older adult reflecting on information about
themselves as they work toward maintaining or improving
their quality of life and independence.
ACKNOWLEDGMENTS
This research was supported by the Aware Home Research
Initiative, Siemens and the Everyday Computing Lab.
Future Field Studies
A considerable amount of effort is needed to conduct a
study as seemingly simple as this one, especially for the
first time. Coordination with one extended family, but two
homes, can be challenging. We were fortunate in that both
Helen and Will were motivated to participate although the
initial motivation came from Will. Installing a network of
sensors in a home is still not routine. Even the pragmatics
of negotiating third-party payment for broadband
connections proved more tiresome than predicted.
REFERENCES
1. Consolvo, S., Roessler, P. & Shelton, B.E. (2004) "The
CareNet Display: Lessons Learned from an In Home
Evaluation of an Ambient Display," Proceedings of the
6th Int'l Conference on Ubiquitous Computing:
UbiComp '04 , (Sep 2004).
2. Consolvo, S., Roessler, P., Shelton, B.E., LaMarca, A.,
Schilit, B., & Bly, S. (2004) "Technology for Care
Networks of Elders," IEEE Pervasive Computing
Mobile and Ubiquitous Systems: Successful Aging, Vol.
3, No. 2, (Apr-Jun 2004), pp.22-29.
With this experience, we have now packaged a more easily
deployable version of the DFP using X10 wireless motion
sensors. The system now includes software for entering a
home floor plan, configuring and calibrating the sensors,
and establishing the connection with the portrait display in
the other home. Part of our future work involves partnering
with other research groups to deploy this technology in
related research trials.
3. Forlizzi, J., SiSalvo, C. & Gemperle, F. “Assistive
Robotics and an Ecology of Elders Living
Independently in Their Homes,” Journal of HCI Special
Issue on Human-Robot Interaction, June 2004.
4. Hirsch, T., Forlizzi, J., Hyder, E., Goetz, J., Stroback, J.,
Kurtz, C. “The ELDer Project: Social and Emotional
Factors in the Design of Eldercare Technologies,”
Conference on Universal Usibility, 2000.
Combining Medical and Activity Monitoring
One such effort is combining our activity monitoring and
wearable medical monitoring for older adults who have
more severe medical concerns. We are working with XXX
to investigate the combination of wearable blood pressure
and glucose level monitors with activity monitoring for
older adults with Type 2 diabetes. Different visualizations
are needed for the clinical physician reviewing data
periodically, for the older adult monitoring his or her own
data, and, for now, the traditional DFP interface augmented
with medical data for family members. We are interested in
how these two types of data aid families in noticing
deviations in behavior and aid clinicians in understanding
the larger context surrounding medical data.
5. Internet Tea Kettle. http://www.mimamori.net/.
6. Morris, M., Lundell, J. & Dishman, E. “Catalyzing
Social Interaction with Ubiquitous Computing: A needs
assessment of elders coping with cognitive decline,”
Proc. Conf. on Human Factors in Computing Systems,
ACM Press, 2004.
7. Mynatt, E.D. and Rowan, J. (2000) “Supporting CrossGenerational Communication” 2000 IFIP HOIT
Conference (HOIT 00).
8. Mynatt, E.D., Rowan, J., Jacobs, A. & Craighill, S.
(2001) “Digital Family Portraits: Supporting Peace of
Mind for Extended Family Members,” Proc. Conf,
UBICOMP, ACM Press, 2004
Activity Monitoring for Proactive Interaction
We have also been approached by another research group
interested in using the DFP technology as a proactive
interface for older adults grappling with debilitating
depression. In addition to providing awareness information
to concerned family members, the system would detect
extended periods of low activity and attempt to encourage
more healthy behavior. A personalized visualization would
also help the older adults assess their good and bad periods,
perhaps helping them draw the correlation between more
9. PhidgetInterfaceKit 0/16/16.
http://www.phidgets.com/index.php?module=pncommer
ce&func=itemview&KID=1095027574199.77.205.234
&IID=57.
10. Shellenbarger, Sue. (2004) “When Elderly Loved Ones
Live Far Away: The Challenge of Long-Distance Care”,
In Wall Street Journal, July 29, 2004
11. Pulsors. http://www.sureaction.com.
530