=Paper=
{{Paper
|id=Vol-3153/paper5
|storemode=property
|title=Using Persuasive Features to Promote Physical Activity for Older Employees – Report from the AgeWell Project
|pdfUrl=https://ceur-ws.org/Vol-3153/paper5.pdf
|volume=Vol-3153
|authors=Niklas Hungerländer,Alessandra Merizzi
|dblpUrl=https://dblp.org/rec/conf/persuasive/HungerlanderM22
}}
==Using Persuasive Features to Promote Physical Activity for Older Employees – Report from the AgeWell Project==
https://ceur-ws.org/Vol-3153/paper5.pdf
Using persuasive features to promote physical activity for
older employees – Report from the AgeWell project
Niklas Hungerländer1[0000-0002-9839-5228], Alessandra Merizzi2[0000-0002-8781-0887] and
Miroslav Sili1[0000-0002-6181-0694]
1
Austrian Institute of Technology, Giefinggasse 4, 1210 Vienna, Austria
2Centre for Socio-Economic Research on Aging, IRCCS INRCA-National Institute of Health
and Science on Aging, Via Santa Margherita 5, 60124 Ancona, Italy
niklas-aron.hungerlaender@ait.ac.at, a.merizzi@inrca.it,
miroslav.sili@ait.ac.at
Abstract. Older people in the cusp of retirement are often concerned about their
future and are worried that retirement means inactivity. The AgeWell project
seeks to support them through their transformation to retirement and beyond. In
this paper the development of the physical activity app is reported. Persuasive
features as well as behavior change techniques were used to digitalize the re-
quirements and wishes of the users for the app. Users tested the app for 10 weeks.
Results indicate that more users were physically active after the testing period
and showed appreciation towards various persuasive aspects of the app. The di-
rection of future research can focus on understanding which persuasive features
are most useful, and testing the level of use of an improved, flexible and intuitive
system.
Keywords: physical activity, elderly, persuasive design, eHealth
1 Introduction
This paper reports1 the application of behavior change techniques to promote physical
activity of older adults in the retirement phase. All work reported happened within the
AAL Joint Program project AgeWell[1]. The AgeWell solution provides activities for
helping older workers and retirees aging healthy during the transition from work to
retirement that can be life change entailing many risks for the physical and mental
health of many older adults e.g., isolation, lack of interests, depression, laziness and
lack of motivation in performing physical activity at detriment of their overall health
and well-being. AgeWell provides an avatar on a mobile phone and robot (physically
embodied device) based virtual coach. The system consists of 5 major components: the
Persuasive 2022, Adjunct Proceedings of the 17th International Conference on Per-
suasive Technology. Copyright © 2022 for this paper by its authors. Use permitted
under Creative Commons License Attribution 4.0 International (CC BY 4.0).
2
avatar, the physical activity reasoner, the mental health reasoner, a medical backend
platform and the mobile robot companion.
Within this paper only the development of the physical activity reasoner and the idea
behind it is reported. The procedure started with gathering insights from focus groups,
followed by a review and selection of suitable items of the behavior change technique
taxonomy (BCT) [2] and the Persuasive Systems Design framework (PSD) [3] which
fit the requests. After conceptualization of the model the physical activity backend was
developed including 20 physical activities. The whole solution was tested within a time
frame of 10 weeks by 62 users, respectively 34 from Italy and 28 from the Netherlands.
Quantitative and qualitative data were collected at three points in time: before the trial
started, at the middle (week number 5) and at the end (week number 10), via online
questionnaires and focus groups.
2 Methods
Since the AgeWell project follows a user-centric approach, focus groups were held with
end-users at the start of the project in the Netherland(N=34), Austria(N=11) and It-
aly(N=10). The results give insights of what is important to users about a virtual coach.
Concerning physical activity following points were stated by the focus group:
1. Autonomy is very important
2. The coach should identify if I set unrealistic goals and warn me, if I train too hard or
too little
3. The coach should have a reminder function and help motivating me towards the goal
In a next step PSD and BCT items where identified which cover the points mentioned
above. The selected items are shown in Table 1. As a result, the following requirements
were made on the physical activity reasoner: It must be aware of the current physical
capability of a user as well as set manageable goals based on previous performances
via adaptive goal setting. The implementation of these two tasks is described in section
2.1. Furthermore, it must give the user autonomy. This is realized by providing the user
multiple physical activities to choose from as well as giving him the choice to choose
the activities he likes and overrule the adaptive goal setting by choosing his own goal
(see section 2.2).
In order to give the users feedback and praise them, a motivational message corpus
which consists of roughly 100 messages was used. Also, self-efficacy is important to
motivate users which was realized with progress bars (see section 2.3).
3
Table 1. Important points stated by the focus group concerning physical activity with fitting
items from the PSD and BCT and implementation.
Item No. BCT PSD Realisation
Focus
Group
1 Action planning Reduction Users can freely choose which ac-
tivities they want to perform when
2 Review behavior goals Personalisation Users receive each week a new
goal which adapts to performance
3 Focus on past success Praise Advise to think about previous
successes
3 Feedback on behavior Self-monitoring use of progress bars
3 Prompts/Cues Reminders Use of reminders for upcoming ac-
tivities
2.1 Identify physical activity capability and setting realistic goals
To be most effective, it is necessary that the prescribed dose of exercises be tailored
to an individual’s current physical capability and desired health goals. In clinical prac-
tice, the FITT-VP principle of exercise prescription is used by experts to tailor physical
activity goals. The principle suggests changing Frequency (f per week), Intensity (i e.g.,
light, moderate, vigorous), Time (t per session), and Type of exercise to adjust or pro-
gress exercise volume (v = i × t × f) per week that correlates with energy expenditure.[4]
At this point Intensity is modelled via the type of exercise in the module (different
exercise types have different intensities). Measures of dose, intensity and physical ca-
pability are central to coaching users in physical activity.
Exercise intensity is described based on energy demands of physical activity. It is
measured through caloric expenditure also known as a MET (metabolic equivalent) that
is defined as the amount of oxygen consumed while sitting at rest. A MET value for
any physical activity is the energy cost expressed as a multiple of the resting metabolic
rate.[5] For example, playing football has more METs than going for a walk. Therefore,
choosing activities with a higher MET value requires less duration to meet the goal than
choosing an activity with a low MET value.
In clinical settings, aerobic capacity is often precisely measured through elaborate
tests of volume of oxygen consumption, or functional evaluations, such as 6-minute
walk tests that are more relevant in deconditioned individuals. Given constraints of an
app-based coach, these objective, clinical evaluations are difficult and costly to imple-
ment. Therefore, a self-report instrument was adapted, the International Physical Ac-
tivity Questionnaire (IPAQ), which is a reliable instrument to assess baseline physical
activity levels [6]. In the questionnaire, users report duration and frequency of physical
activities in a typical week. These questions probe duration and frequency for low in-
tensity (3 METs, such as stretching), moderate intensity (5 METs, such as fast walking),
and vigorous activities (8 METs, such as playing football). Given these measures, an
4
individual’s initial physical capability (in METs) can be computed. The implementation
of the IPAQ is depicted in Fig. 1.
Apart from having an intensity measure with using MET values it was also used to
motivate users, by providing them weekly credits equivalent to the MET values. With
the abstraction from a time unit (e.g. 150 minutes) to a credit (e.g. 750 Credits) a gam-
ification technique is used as well as users get driven away from focusing and worrying
on how much time they must spend to reach their goal.
Fig. 1. Implementation of the IPAQ questionnaire
Beside setting a proper initial goal the fitness coach must be aware of the weekly per-
formance and must set new manageable goals based on previous performances. Since
no guidelines of how to increase/decrease physical activity based on previous perfor-
mance are available in literature a handwritten rule is applied. Physical activity is in-
creased when users perform more than 75% of their last goal. Specifically, for every
percent higher than 75% their next goal is increased by 1% of the last goal. For example,
a user which achieved 90% of his last goal gets an increase of 15% for the next goal.
On the contrary, activity is decreased if users perform less than 75% of their last goal.
However, for every percent lower than 75% their next goal is decreased by 0.33% of
the last goal. As a result, the increase is much steeper than the decrease. The idea behind
this application is that according to Kahnemann and Tversky’s prospect theory, losses
are weighting heavier than gains and should therefore be smaller to keep up motivation
[7].
2.2 Ensuring autonomy
After users received their first weekly credits, they could freely choose between 20
activities (e.g. walking, swimming, gymnastics) they like to perform during the week.
It is important to make the goal as precise as possible to raise the chance of completion
(on which day, at which time, for how long) but at the same time give the users the
autonomy the requested. Therefore, users planned only the type and duration of an ac-
tivity and always had the chance to adapt it. The implementation is shown in Fig. 2.
5
Fig. 2. Implementation of choosing an activity. Users could select the type of activity as well as
the day(s) and duration. At the last screen they received a summary telling them how many
credits they had left for the week. In this example, the user planned more activities then the
coach has planned for him and therefore he gets an advice to not overdo it.
After an activity has been scheduled for a specific day, the user is asked at a predefined
time if the activity was performed, as well as for specific feedback.
Although Intensity is implemented in the current module via different physical ac-
tivities (METs), it also plays a role when asking users for feedback after activities. The
Borg’s RPE scale[8] was used to ask the exertion level of users on three levels, no –
moderate – high exertion.
The reasoner chooses dependent on the answer a proper message. It will be positive
in all cases since the user managed to do the activity. Furthermore, if he chose “no
exertion” he additionally receives a message to increase the intensity the next time. If
he chose “high exertion” he additionally receives a message to decrease the intensity
the next time (see Table 2).
In case the user has not finished his activity, he is asked to provide a reason. Possible
answers are:
• “It was too hard”
• “I had no time”
• “I had no motivation”
• “I don’t think this is useful”
The reasoner chooses a proper message depending on the answer. If the user chose “it
was too hard” he receives a neutral message as well as a message to decrease the inten-
sity the next time (see Table 2). If he chose “I had no time” he receives a negative
message as well as a message which motivates him to stick to the goal. If he chose “I
had no motivation” he receives a message which should motivate him to do better the
next time. If he chose “I don’t think this is useful” he receives a message which provides
information highlighting the benefits of physical activity. The implementation of an-
swering to a scheduled activity is depicted in Error! Reference source not found..
6
Fig. 3. Two possible outcomes of answering questions after a scheduled activity.
After a week is over the user receives a message including the credits he was able to
achieve and get a new credit suggestion dependent on how active he was (see section
2.1 for details). However, to ensure user autonomy users can increase or decrease the
credits for the upcoming week. The implementation is depicted in Fig. 4.
Fig. 4. Message at the end of the week. Information on how many credits were achieved is pro-
vided. A new credit suggestion for the upcoming week is included which can be adapted by the
user.
2.3 Remind and motivate
Throughout the week the user receives every day a task is scheduled a reminder at a
preferred time specified by the user in the settings. The reminder includes information
on the type of activity and duration as well as motivation based on the feedback from
the last time. Specifically, if the last activity was indicated as too easy a message is
included which encourages the user to increase the intensity this time. If the activity
was indicated as too hard the opposite is the case. (see Table 2). The user can skip,
7
postpone or take note of the message. The implementation of the reminder is depicted
in Error! Reference source not found..
Fig. 5. Reminder for an upcoming activity.
For creation of the motivation message corpus the work of Vries et al[9] was used as
starting point. They created an extensive corpus of messages via crowdsourcing which
had the purpose to motivate for exercise adherence. A sub sample of this work was used
in the AgeWell project, some of the messages were adapted and new messages were
created. All messages were annotated to be ready for use by the reasoner.
Table 2. Types of motivational messages and example texts
Option Example
Positive You are on the right path! Stay there and hold on!
Increase next time Next time you exercise, plan on pushing yourself a bit harder.
Decrease next time You did great! I think it's important to keep in mind that you don't
want to overdo it and hurt yourself though!
Neutral Remember, changing behavior takes time. Don't be too hard on
yourself if it sometimes doesn't work. It will get better next time.
Negative Wouldn't it be worth for your own well-being to start exercising?
Motivation Take care of yourself so your health doesn't become a burden on
other people.
Information Even small amounts of exercise every day are shown to significantly
reduce the risk of heart disease
Increase before activity Last time you did a lot, but I know you have it in you to increase the
intensity.
Decrease before acti- After that great exercise the other day let's focus on keeping your in-
vity tensity a little lower.
Positive weekly You made it through the week! That wasn't so hard, was it? Keep
up the good work!
Negative weekly You didn't reach the goal this time, but it's all about consistency.
Keep exercising!
Motivation weekly You've put so much work into this thus far, it doesn't make sense not
to keep it up.
8
In order to give users feedback on their activities progress bars are used (see Fig. 6).
Users can watch their performance in the last five weeks and if they reached their
weekly goal. Additionally, textual information is provided which puts focus on the pos-
itive aspects, motivating users.
Fig. 6. Progress bar to boost self-efficacy with textual information and motivation.
3 Results
For the analysis app usage data and data from the three questionnaires and online meet-
ings were used. Data from the questionnaires and focus groups were collected before
the trial started, at the middle (week number 5) and at the end (week number 10) of the
trials. Out of the 62 users 42 used at least one week the physical activity section in the
app and were therefore included in the app analysis. 35 users completed all three ques-
tionnaires and were therefore included in the questionnaire analysis.
3.1 Results from the questionnaires
The questionnaires included quantitative questions with space for qualitative com-
ments. Users were asked how active they felt on a 3-point scale. At the baseline 50.8%
of them felt active or very active, while at the end of the trial 56.3% did.
The questionnaires also included the IPAQ, providing a quantitative measure for
activity. Interestingly, users were significantly more days active in week number five
than at the start of the trial (P=.027). However, there was no significant difference when
comparing active days at the start of the trial with week 10 or active days in week 5
compared with week 10.
3.2 Results from the online group meetings
Participants from both countries expressed appreciation for the usefulness of the system
for motivating the practice of healthy activities, including physical ones, and for
9
promoting active ageing attitudes and behavior. Particularly, the Italian participants
recognized:
- the power of the system to motivate people to increase physical activity and
to do it more regularly;
- the usefulness of receiving reminders to carry out activities suggested by the
avatar;
- the importance of the app in helping users focus on their needs, including the
physical ones.
3.3 Results from the app data
While using the app, users were asked for every scheduled activity if the duration of
the activity they choose was either right or too short/long. Furthermore, they could in-
dicate if the activity was too easy or too hard or reasons why they didn’t do an activity.
1033 of scheduled activities were indicated as completed. In more than half of the cases
the activity duration and intensity was just right. 2 times the activity was indicated as
too hard and 335 times the activity was indicated as too easy.
66 times an activity was indicated as not completed, of which 2 times because of no
motivation, 59 times because of not enough time and 5 times because the activity was
not seen as useful to the user (see Table 3).
Table 3. Chosen answers to the question if a scheduled activity was done and answers to follow-
up questions
Activity dura- too hard too no motiva- no time not use-
tion was good easy tion ful
Activity Done 696 2 335
Activity not Done 2 59 5
4 Discussion
This paper reported on the development of a physical activity app which includes the
requirements and wishes of users via persuasive features and behavior change tech-
niques.
4.1 Findings
Results show some positive outcomes from the use of the system for scheduling and
performing physical activities. Participants reported positive feedback on the way the
app motivated them with scheduling and carrying out physical activities, and in fact it
was possible to observe their engagement with such activities through the analysis of
quantitative data from the app itself. The increasing trend observed in the initial 5 weeks
may have stabilized for the following 5 weeks for several reasons:
10
• they found their balance and did not want to increase their activities further
• they had problems remembering to record the activity as completed
• they fell ill or needed to drop their commitment slightly to care for someone else
(impact of Covid, over 20% of participants infected throughout the trial).
Overall, the AgeWell app system seems promising, and with some adaptations can be
a good solution for encouraging retirees to keep or become physically active.
4.2 Research added value
Given the lack of research regarding the promotion of physical activity for older
adults[10] it is difficult to gather promising features from literature. Previous research
already showed that co-design of fitness apps with the intended user, namely older
adults, has a positive effect on technology experience[11]. Similarly, this paper high-
lights the benefits of adopting a user-centered design approach to select persuasive fea-
tures and behavior change techniques. This brings advantages over choosing the fea-
tures solely from theory or even randomly due to the lack of clear guideline on which
feature works under which contexts. The approach presented here introduces a tech-
nique to gather the most important features for a specific target group and use-case
when data on effectiveness of persuasive features and behavior change techniques is
sparse.
4.3 Limitations
A limitation of the study is that participants were not asked to reflect on their process
of change when using the app, i.e., whether they found it more helpful in changing their
behavior to receive reminders or to view the progress bar graph. Although participants
gave qualitative feedback on these two aspects, accurate quantitative data would have
helped to observe whether they preferred one over the other.
Another limitation concerns the usability of the app was not so well perceived as
some technical aspects need improvement, including its intuitiveness and flexibility in
planning and recording physical activities. The findings would have probably been
more positive if the system was further developed prior testing. However, a product
cannot be finalized before it is tested, so feedback from participants will be useful to
refine the product.
5 Conclusion
This paper presents a solution to use persuasive system design principles and behavior
change techniques to motivate older adults in the retirement phase doing physical ac-
tivity. It adopted a user-centered design approach to gather the most important features
the solution should contain according to end-users and experts.
It was then tested on a sample of retirees and older workers who showed appreciation
towards various persuasive aspects of the app which was confirmed by the number of
11
reported physical activities, particularly during the initial 5 weeks. The direction of fu-
ture research can focus on understanding which persuasive features are most useful,
and testing the level of use of an improved, flexible and intuitive system.
Acknowledgements
This research was co-funded by Active and Assisted Living Program (reference no. aal-
2018-5-92-CP).
References
1. AgeWell. https://agewell-project.eu/
2. Michie S, Richardson M, Johnston M, et al (2013) The behavior change technique
taxonomy (v1) of 93 hierarchically clustered techniques: Building an international
consensus for the reporting of behavior change interventions. Ann Behav Med 46:81–
95. https://doi.org/10.1007/s12160-013-9486-6
3. Harri O-K, Marja H (2009) Persuasive Systems Design: Key Issues, Process Model, and
System Features. Commun Assoc Inf Syst 24:96.
https://doi.org/10.17705/1CAIS.02428
4. Medicine AC of S (2013) ACSM’s guidelines for exercise testing and prescription.
Lippincott Williams & Wilkins
5. Jetté M, Sidney K, Blümchen G (1990) Metabolic equivalents (METS) in exercise
testing, exercise prescription, and evaluation of functional capacity. Clin Cardiol
13:555–565
6. Lee PH, Macfarlane DJ, Lam TH, Stewart SM (2011) Validity of the international
physical activity questionnaire short form (IPAQ-SF): A systematic review. Int J Behav
Nutr Phys Act 8:1–11
7. Kahneman D, Tversky A (2013) Prospect theory: An analysis of decision under risk. In:
Handbook of the fundamentals of financial decision making: Part I. World Scientific,
pp 99–127
8. Williams N (2017) The Borg rating of perceived exertion (RPE) scale. Occup Med (Chic
Ill) 67:404–405
9. De Vries RAJ, Truong KP, Kwint S, et al (2016) Crowd-Designed Motivation:
Motivational messages for exercise adherence based on behavior change theory. In:
Proceedings of the 2016 CHI conference on human factors in computing systems. pp
297–308
10. Aldenaini N, Alqahtani F, Orji R, Sampalli S (2020) Trends in Persuasive Technologies
for Physical Activity and Sedentary Behavior: A Systematic Review. Front. Artif. Intell.
3
11. Harrington CN, Wilcox L, Connelly K, et al (2018) Designing health and fitness apps
with older adults: Examining the value of experience-based co-design. In: ACM
International Conference Proceeding Series. pp 15–24