Using Mobile & Personal Sensing Technologies to Support

Health Behavior Change in Everyday Life: Lessons Learned

Predrag Klasnja1, Sunny Consolvo1,2, PhD,

David W. McDonald1, PhD, James A. Landay1,2, PhD, & Wanda Pratt1, PhD
1
University of Washington, Seattle, WA; 2Intel Research Seattle, Seattle, WA
Abstract
Lifestyle modification is a key facet of the prevention
and management of chronic diseases. Mobile devices
that people already carry provide a promising
platform for facilitating these lifestyle changes. This
paper describes key lessons learned from the
development and evaluation of two mobile systems
for encouraging physical activity. We argue that by
supporting persistent cognitive activation of health
goals, encouraging an extensive range of relevant
healthy behaviors, focusing on long-term patterns of
activity, and facilitating social support as an optional
but not primary motivator, systems can be developed
that effectively motivate behavior change and provide
support when and where people make decisions that
affect their health.
Introduction
Although pharmacological advances have made great Figure 1: Houston (left) and UbiFit (right)
strides in decreasing morbidity and mortality from we have conducted early stage field studies of mobile
chronic diseases, lifestyle modification remains a key technologies designed to encourage physical activity.
aspect of effective chronic disease management. In this paper, we describe key lessons learned from
Interventions that target lifestyle modification have that work in an effort to help others who are
been shown to be effective in the prevention and designing systems to support health behavior change.
management of heart disease,1 diabetes2 and obesity.3 We conclude with methodological reflections about
Yet, patient compliance with lifestyle modification how to design such systems so that they smoothly
remains low. For example, fewer than half of heart integrate into users’ everyday lives while effectively
disease patients continue to exercise six months encouraging lifestyle change.
following the completion of cardiac rehabilitation.4
Systems and Field Studies
Numbers are similar for compliance with dietary
recommendations.5 Why is behavior change so We have designed two mobile phone-based systems
difficult to achieve? Simply put, it is a complex to encourage regular physical activity: Houston and
process. Even a single change, such as increasing UbiFit. Houston8 (Figure 1, left), the first system we
physical activity, likely requires the individual to designed, uses a mobile phone application and a
restructure her priorities as well as her daily and pedometer to encourage users to increase their daily
social routines, such as finding time for exercise in step count. The phone application provides a journal
the midst of work and family obligations. where users can review trends of their daily step
Encouraging health-promoting lifestyle change counts, add comments to their step counts, receive
requires that interventions be integrated into small rewards for reaching their daily goal, share
everyday life, with support available when and where their step counts with ‘fitness buddies,’ and exchange
individuals make decisions that affect their health. messages with those buddies. We conducted a three-
week field study of Houston with 13 participants,
Mobile technologies that individuals routinely carry,
comprised of three groups of female friends from
such as mobile phones, may be a particularly
pre-existing social networks. Each participant was
effective platform for delivering such encouragement
interviewed at the beginning of the study, after the
as they are likely to be with the individual when she
first week, and at the end of the study.
most needs the support.6,7 Over the past several years,

AMIA 2009 Symposium Proceedings Page - 338

Based on our experiences with Houston, results from sensing device (full system). Each participant was
other persuasive technology research, and behavioral interviewed at the beginning of the study, at the end
and social psychological theories,9 we designed of the first month, and at the end of the study.
UbiFit10,11 a system that uses a mobile phone and a
In the following sections, we discuss key lessons
sensing device to encourage regular and varied
learned from these two research projects.
physical activity. Two of UbiFit’s components run on
the user’s mobile phone: (1) an interactive Lessons Learned
application used to journal physical activities, review
Four lessons from our work are particularly relevant
activities done on any given day, and track progress
to the design of mobile systems for encouraging
toward a weekly goal, and (2) a glanceable display
health behavior change. These are the importance of
that uses an abstract, stylistic representation of the
(1) supporting persistent cognitive activation of
physical activities the user performs each week,
health goals, (2) encouraging an extensive range of
displayed on her phone’s background screen. The
relevant healthy behaviors, (3) focusing on longer-
glanceable display provides weekly goal attainment
term patterns of activity, and (4) facilitating social
status, physical activity behavior, and a subtle,
support as an optional but not primary motivator.
persistent reminder of her commitment to physical
activity. In our implementation, the display uses a Supporting Persistent Activation of Health Goals
garden metaphor to represent a week’s worth of
physical activity behavior. The garden blooms with An important result of the three-month experiment of
different types of flowers to represent the different UbiFit was the significant difference in the weekly
types of activities the user performs: walking, cardio, activity level between participants who had and did
strength, flexibility, and other non-exercise physical not have the glanceable display. The former group
activities (e.g., housework). Upon reaching her maintained their activity levels throughout the study,
weekly goal, a large butterfly appears. Up to three even though the winter holiday season is known for
smaller butterflies represent goal attainments for the physical inactivity. In contrast, the activity levels of
prior three weeks (Figure 1, right). the latter group decreased significantly. This result is
consistent with the social psychological literature13
In addition to the mobile phone components, UbiFit on automatic goal activation. This literature finds that
uses the Mobile Sensing Platform (MSP),12 a pager- goals can be activated through environmental cues,
sized, battery-powered computer that uses a and that such “primed” goals can effectively guide
barometer and three-dimensional accelerometer to goal-directed behavior. The glanceable display kept
automatically detect the duration and start time of the physical activity goals chronically activated,
walking, running, cycling, stair machine, and enabling participants who had the display to stay
elliptical trainer activities. When the MSP is worn on engaged with their commitment to physical activity.
the waistband and within Bluetooth range of the A participant explained:
phone, these activities are detected automatically. As
they are detected, the activities appear both in the [The garden] was a constant reminder…whereas if
interactive application and on the glanceable display. you didn’t have a [garden], you probably—I wouldn’t
think about [physical activity] as much, you know.
We conducted two field studies of the UbiFit system: [With the garden] I think about it maybe
a three-week trial and a three-month experiment. In subconsciously every time I look at my phone.
the three-week trial,10 12 participants used UbiFit and
provided feedback on the system and the automatic Supporting this kind of persistent activation of health
activity detection. As with the Houston study, each goals can be a powerful means of fostering health
participant was interviewed at the beginning of the behavior change. Although a number of commercial
study, after the first week, and at the end of the study. mobile phone applications enable users to track their
Based on the results, we redesigned elements of the diet and physical activity, they are likely to yield
system, including improving the activity detection. results similar to those of our no glanceable display
condition because they do not provide the persistent
The revised version of UbiFit was evaluated over the visual reminder. UbiFit is, to our knowledge, the only
winter holiday season in a three-month experiment health application to date to use the phone’s
with 28 participants.10 Participants were randomly background screen to provide users with continuous
assigned to one of three experimental conditions: feedback about a behavior they are trying to change.
interactive application and sensing device only (no In addition, the stylized nature of UbiFit’s display
glanceable display), interactive application and allowed users to maintain some level of privacy
glanceable display only (no sensing device), or should their phone be seen by someone else.
interactive application, glanceable display, and

AMIA 2009 Symposium Proceedings Page - 339

Future directions. UbiFit’s glanceable display subtly the three-month experiment, including skiing, cardio
reminded participants about their commitment to classes, dancing, swimming, and ice skating.
physical activity and it provided feedback about their
This experience highlights an important lesson that
recent activities and goal attainment. Future work
needs to be considered when designing systems for
should investigate whether the former is effective
the support of health behavior change. Such systems
without the latter. Other types of representations, for
not only help users track and modify their behavior,
example, providing an encouraging message or
but insofar as the user becomes invested in using the
drawing from a loved one, could be explored.
system, the system also shapes how she thinks about
Assuming the user strongly associates the
the behaviors she is trying to change. The type of
representation with her health goals, it could be
credit that the system provides could inadvertently
effective at achieving persistent goal activation.
encourage the user to focus only on activities that the
Whether such representations are sufficient on their
system supports, potentially at the expense of
own or if explicit, persistent feedback about recent
activities that might be, from a health perspective,
activities is necessary should be further examined.
equivalent or even more important.
Encouraging an extensive range of healthy behaviors
Future directions. With Houston, we tracked step
Our work suggests that the system can substantially count only; with UbiFit, we tracked and encouraged
influence how the individual engages in health the range of relevant physical activities. To continue
behavior change. Specifically, the activities that the this trajectory, the range of healthy behaviors that are
system supports or encourages can become the focus encouraged by the system could be further expanded,
of the user’s efforts, potentially to the exclusion of especially when the system is targeting the
other relevant activities. prevention and management of chronic diseases. In
the case of heart disease, for example, a patient might
An example will help clarify this point. Following
not only need to increase physical activity, but also
health science literature on the effectiveness of
change her diet, reduce stress, and stop smoking. Our
pedometers, Houston attempted to encourage
findings suggest that an effective system will support
physical activity by helping users track their daily
an extensive range of the healthy behaviors within
step count. Users could add comments to their step
the relevant areas of lifestyle change.
count (e.g., “Went for a bike ride”), but the system
provided no other explicit functionality for tracking However, it is unlikely that every user will need to
other forms of physical activity. This resulted in an focus on changing all of those aspects of her life, or
unintended, negative side-effect. Several participants at least not all at the same time. Providing
realized that the pedometer did not capture cardio customization that allows the user (or health care
activity well—for example, running three miles provider) to select aspects of the system that are
yields a lower step count than walking the same three appropriate for the user’s current needs, and adjust
miles and cycling yields no steps at all. As a the system as her needs change, may improve the
frustrated participant explained, “my main source of effectiveness of the system over time.
exercise [rock climbing] doesn’t register.” Similarly,
Focusing on long-term patterns of activity
the pedometer did not distinguish between steps
made while walking on a flat surface or up hills— Behavioral economics claims that individual actions
although these activities differ in their intensity and can have a very different value than the patterns of
in their ability to help individuals lose weight. those same actions.14 If, while on her morning coffee
run, an individual is deciding between ordering a
Because the system did not provide proper credit for
black coffee (0 calories) and a caramel frappuccino
these types of exercise, several participants simply
(380 calories), the tasty frappuccino might appear
chose not to do them. A participant noted, for
much more appealing. However, if she is deciding
example, that the pedometer did not “care whether
which of those beverages to have every morning of
you went up and down hills or whether you walk on
the week, she may decide that the 2660 calories from
flats, so why kill yourself?” This outcome was not
seven frappuccinos are not worth it. The difficulty is
what we intended. Based on this experience, we
that in the moment that decisions are made,
trained UbiFit’s sensing device to detect a wider
individuals tend to focus on the current decision, and
range of activities (walking, running, cycling,
not on the pattern that such decisions form over time.
elliptical trainer, and stair machine), and we allowed
participants to journal any other physical activity in With UbiFit, the week’s worth of activities and
the interactive application. As a result, 26 types of month’s worth of goal attainments represented on the
cardio activities were performed by participants in glanceable display encouraged participants to think
about physical activity not as a one-off choice (e.g.,

AMIA 2009 Symposium Proceedings Page - 340

Do I need to work out today?) but rather in terms of when they were less active, and some felt that sharing
patterns of behavior (e.g., What did I do last week? introduced too much competition. The effect of social
What have I done so far this week? What can I still support on users’ motivation was decidedly mixed.
do to have an active week?).
Based on these findings and similar results from
Helping participants reflect on a week’s rather than a others,15 we realized that while social support can be
day’s worth of activity in the display meant that even helpful, it should not be the primary strategy used to
if the participant had a couple of sedentary days, she motivate health behavior change. With UbiFit, we
would not necessarily be discouraged, as she could designed a system to motivate users to engage in
still have a good week. Just as importantly, seeing a physical activity without using social support as a
week’s worth of activity helped many realize how primary motivator. However, sharing regularly took
inactive they were—this awareness surprised most place anyway. Participants routinely showed their
participants—and take concrete steps to be more gardens to family and friends, and for some, family
active. A participant explained: members helped to encourage physical activity. A
participant explained,
I used [the glanceable display] to increase my
awareness of what I was doing…’cause like…after [My daughter] would really encourage me to [be
about two days, you kind of forget, like ‘did I really active] and she would ask me for pink flowers all the
do that or am I just dreaming or was that last week?’ time...She was very excited, and she wanted [me to
get] the butterflies.
Encouraging users to reflect on how each choice they
make forms a pattern of behavior over time can be a Similarly, family members and friends who
powerful way to encourage health behavior change. participated in the study together often compared
Mobile devices offer an advantage over Web-based their gardens, somewhat replicating the sharing
tools as they are often with the user when decisions functionality explicitly facilitated by Houston. As
are being made. A participant explained: with Houston, however, UbiFit participants felt
uncomfortable when someone would ask to see their
I liked having [my garden display] be on the
garden when they had not been very active.
phone…something I have with me…[with] a Web site,
it’s so easy, ‘oh, I didn’t do anything, I’m not going Future directions. While social support can be a
to click on it.’ It’s so easy to ignore it. But on the powerful strategy for encouraging behavior change—
phone, you can’t really ignore it as and users engage in it even when the system does not
easily…Otherwise, it’s just…out of sight, out of mind. facilitate it—it can also hamper motivation and even
introduce social friction (if, for example, someone
Over time, such tools could help users learn to
initially shares her data, then later decides to stop).
conceptualize choices naturally in terms of how they
Although health behavior change systems could
create patterns, thus helping users internalize one of
benefit from facilitating social support, they should
the most powerful means of ensuring self-control.14
not depend on it. In addition, future work should
Future directions. Future systems could further explore how to incorporate social support. For
explore supporting reflections on patterns of example, while the system should allow users to
behavior. For example, a system could help the user determine what to share with whom, an open design
realize how other factors affect her healthy and challenge is how to allow the user to adjust these
unhealthy activities (e.g., location or other people). settings easily as circumstances change. Without such
Similarly, a context-aware mobile tool could learn control, social support can backfire, ultimately
over time what challenges the user faces in trying to leading to system abandonment.
lead a healthy lifestyle and proactively provide
Discussion & Conclusion
support when she most needs it.
Unlike technologies that individuals have to use—as
Facilitating but not depending on social support
part of their jobs, for example—the use of
Houston explicitly facilitated social support by technologies for health behavior change is often
providing features that enabled users to share step discretionary. For such systems to be effective and
counts, goal progress, and messages with their fitness continue to be used, they must be well designed with
buddies. What we discovered in the Houston study, careful consideration given to how they will fit into
however, was that social support was a double-edged everyday life. How functionality such as journaling,
sword: participants enjoyed sharing their step count feedback, or social support is designed can make or
when they were being active and were motivated by break the effectiveness of a system, and even lead to
positive feedback and seeing their buddies do well. system abandonment (and, potentially, abandonment
But participants were often not comfortable sharing of the behavior the system was trying to encourage).

AMIA 2009 Symposium Proceedings Page - 341

Designing for integration into everyday life while analysis vs. prescriptive diet. Nutrition, metabolism,
effectively encouraging health behavior change is and cardiovascular diseases. 2008.
complicated. Important effects of a system’s design 3. Galani C, Schneider H. Prevention and treatment
often cannot be anticipated in advance. When a of obesity with lifestyle interventions: review and
system is used in the field by individuals from the meta-analysis. International journal of public health.
target audience, even for a short period of time (e.g., 2007;52(6):348-59.
a few weeks), important issues surface that cannot 4. Moore SM, Dolansky MA, Ruland CM,
easily be predicted or found during initial system Pashkow FJ, Blackburn GG. Predictors of women's
design or in usability lab evaluations. exercise maintenance after cardiac rehabilitation.
Journal of Cardiopulmonary Rehabilitation.
In our work, we follow a user-centered design
2003;23(1):40-9.
process, common to the field of human-computer
5. Twardella D, Merx H, Hahmann H, Wüsten B,
interaction (HCI). We engage target users early and
Rothenbacher D, Brenner H. Long term adherence to
often. We employ surveys, as well as beta testing
dietary recommendations after inpatient
with the research team and friends and family for
rehabilitation: Prospective follow up study of patients
weeks to months. Following beta testing, we employ
with coronary heart disease Heart. 2006;92:635-40.
short-term field studies (e.g., a few weeks) with small
6. Fogg BJ. Mobile persuasion: 20 perspectives of
numbers of participants (e.g., 12-15) who are asked
the future of behavior change. Palo Alto, CA:
to use the technology in their everyday lives. We
Stanford Captology Media; 2007.
redesign our systems based on these results prior to
7. Tufano JT, Karras BT. Mobile eHealth
going into the field with more formal, long-term
interventions for obesity: A timely opportunity to
experiments. In our studies, we combine quantitative
leverage convergence tools. Journal of Medical
data collection methods with qualitative methods,
Internet Research. 2005;7(5):e58.
where participants are often interviewed in depth
8. Consolvo S, Everitt KM, Smith I, Landay JA.
about their experiences with the system. The insights
Design requirements for technologies that encourage
we have gained from the qualitative methods have
physical activity. CHI 2006; 2006. p. 457-66.
been critical to our developing a system that has
9. Consolvo S, McDonald DW, Landay JA.
effectively encouraged health behavior change.
Theory-driven design strategies for technologies that
Such HCI-style practices could greatly improve the support behavior change in everyday life. CHI 2009.
effectiveness of technologies developed for medical 10. Consolvo S, Klasnja P, McDonald DW, et al.
informatics research. It is through rich qualitative Flowers or a robot army? Encouraging awareness &
data and use in the field that subtle, yet critical, activity with personal, mobile displays. UbiComp
design problems are often revealed, the timely '08; 2008. p. 54-63.
discovery of which may determine how effective a 11. Consolvo S, McDonald DW, Toscos T, et al.
system proves to be in clinical trials. (See 16 for a Activity sensing in the wild: a field trial of UbiFit
discussion of the value of such data even in RCTs) Garden. CHI '08; 2008. p. 1797-806.
12. Choudhury T, Borriello G, Consolvo S, et al.
As the prevalence and cost of chronic diseases
The Mobile Sensing Platform: An Embedded
continue to rise, the need for lifestyle modification as
Activity Recognition System. Pervasive Computing,
a means of prevention and treatment becomes
IEEE. 2008;7(2):32 - 41.
greater. In this paper, we have argued that carefully
13. Fitzsimons GM, Bargh JA. Automatic self-
designed mobile interventions can be a powerful way regulation. In: Vohs KD, Baumeister RF, editors.
of fostering health behavior change. By supporting Handbook of self-regulation: Research, theory, and
the persistent activation of health goals, encouraging
applications. New York: Guilford; 2004. p. 151-70.
an extensive range of relevant healthy behaviors,
14. Rachlin H. The science of self-control.
focusing on patterns of activity, and facilitating
Cambridge, MA: Harvard University Press; 2000.
optional social support, effective systems can be
15. Lin JL, Mamykina L, Lindtner S, Delajoux G,
designed to help people live long, healthy lives.
Strub HB. Fish'n'Steps: Encouraging physical activity
References with an interactive computer game. Ubicomp 2006;
2006: Springer; 2006. p. 261-78.
1. Zafari AM, Wenger NK. Secondary prevention 16. Hurling R, Catt M, De Boni M, et al. Using
of coronary heart disease. Archives of physical internet and mobile phone technology to deliver an
medicine and rehabilitation. 1998;79(8):1006-17. automated physical activity program: Randomized
2. Forlani G, Lorusso C, Moscatiello S, et al. Are controlled trial. Journal of Medical Internet Research.
behavioural approaches feasible and effective in the 2007;9:e7.
treatment of type 2 diabetes? A propensity score

AMIA 2009 Symposium Proceedings Page - 342