User-Perception of a Webcam-Based Intervention System for

Healthy Habits at Computer Workstations

Angelina Clara Schmidt

, Dimitri Kraft

, Fabienne Lambusch

and Michael Fellmann

Institute of Business Informatics, University of Rostock, Rostock, Germany

Fraunhofer IGD, Rostock, Germany

Keywords: Webcam-Based Intervention System, Healthy Habits, Computer-Based Work.

Abstract: Fueled by the ongoing digitalization, the amount of computer-based work is on the rise. Employees increas-

ingly spend large parts of their day in front of computer workstations. While this type of work means less

physical effort, it can nevertheless cause a range of health problems such as eye strain, back pain, wrist pain,

and muscle fatigue and, in the long run, can lead to serious problems. Although some monitoring systems for

health-related parameters have been developed so far, few of them provide interventions during work. Also,

empirical insights on how users actually perceive such systems are still missing. Hence in our work, we report

on first results regarding the user perception of such systems based on CareCam, a webcam-based system for

health-promoting interventions. Based on user feedback from real-world usage of the system for one week,

we derive insights for the further development of such systems.

1 INTRODUCTION

In many industries, machines have taken over the re-

petitive, physically strenuous and dangerous work

while humans are still needed to perform administra-

tive, management or creative tasks. In line with this,

a proliferation of office work and sedentary work

styles can be observed on a global scale (Park et al.,

2020). The number of computer terminals in work-

places in Germany has even grown by almost 50%

within the last decade (Fichter et al., 2012). In line

with this, 40% of the employees in Germany are

knowledge workers (Burkhart & Hanser, 2018) and

half of the working population is using computers for

work tasks (Bitkom, 2018). However, extensive com-

puter-based work in conjunction with unfavorable

work behaviors can lead to serious health problems.

The most common computer-related health problems

include visual problems such as eye strain and mus-

culoskeletal problems such as back pain, wrist pain

and muscle fatigue (Mary & Munipriya, 2011). Stress

and headaches can also be triggered. These health

problems of employees can even result in reduced

https://orcid.org/0000-0002-6967-4287

https://orcid.org/0000-0002-0604-5854

https://orcid.org/0000-0002-0303-1430

https://orcid.org/0000-0003-0593-4956

productivity, prolonged absenteeism, and early retire-

ment. Therefore, preventive and health-promoting

support in the workplace can play an essential role in

maintaining employees’ health. It moreover offers the

potential to take appropriate action in the early stages

(Mary & Munipriya, 2011). Providing tools for man-

aging and improving health can contribute to the

companies’ competitiveness and could even help to

attract new employees.

With CareCam, an application that can record var-

ious health-related data via a simple webcam has been

developed at Fraunhofer. This data can be used to

identify health strains in the workplace and generate

personalized health-promoting interventions. Never-

theless, it is unclear so far how such interventions can

best minimize health risks at computer workstations

and how users perceive them. Hence, this work aims

to evaluate a concept for health-promoting interven-

tions at computer workstations that has been inte-

grated into CareCam. While the interventions have al-

ready been described from a more technological per-

spective in our previous work (Kraft et al., 2022), we

here put emphasis on user perceptions regarding

Schmidt, A., Kraft, D., Lambusch, F. and Fellmann, M.

User-Perception of a Webcam-Based Intervention System for Healthy Habits at Computer Workstations.

DOI: 10.5220/0011776100003414

In Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) - Volume 5: HEALTHINF, pages 567-580

ISBN: 978-989-758-631-6; ISSN: 2184-4305

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

567

these. To do so, we conducted a pilot study in which

test users employed CareCam with the interventions

in their everyday work environment. Based on semi-

structured interviews, we provide insights into how

users perceived different types of camera-based

health-promoting interventions and if features were

missing. We moreover also derive concrete possibili-

ties for the improvement of future health-promotion

systems for high-screen-time work.

The next section describes related work. Section 3

presents basic features of CareCam while Section 4

focuses on the interventions implemented. The case

study preliminary results are presented in Section 5.

2 RELATED WORK

Most works in the area of camera-based health sys-

tems at computer workstations address ergonomics

and focus on the detection and prevention of un-

healthy sitting postures and durations (Herrera et al.,

2021; Mary & Munipriya, 2011; Paliyawan et al.,

2014 - 2014a, 2014 - 2014b; Taieb-Maimon et al.,

2012) (for a review see (Kraft et al., 2022)). (C. Chen

et al., 2012) present a rather comprehensive system,

which uses a webcam and additional cameras at the

workstation to provide feedback on the current ergo-

nomic state of the worker. For this, parameters such

as average work and break times, distance to the com-

puter screen, head movements, gaze directions, and

blink frequencies are recorded. The system learns the

user’s behavior patterns such as “close to the screen”,

“low head mobility”, or “absent” and provides ergo-

nomic reminders. Beyond sitting postures, some sys-

tems also provide features for advanced monitoring

of health-related parameters such as heart rate, stress,

or mood. For example, (Maeda et al., 2016) capture

physiological information through the normal equip-

ment of a computer workstation. In this process, the

camera measures vital data such as heart rate, facial

expressions, and eye blinks in order to present it in a

visualization application. Another system in this di-

rection has been developed by (Vildjiounaite et al.,

2018), which can recognize stressful working hours.

Finally, there are a few systems that provide health-

promoting interventions. In this direction, (Taieb-

Maimon et al., 2012) present an automatic feedback

system that displays webcam photos of the current

posture at work contrasted with photos of the correct

posture, thus striving to continuously urge the user to-

wards improving his or her posture. Also, (Herrera et

al., 2021) have already introduced active breaks with

exercises, but these are also only related to posture.

The focus is on the automated control of the correct

performance of these posture-related exercises.

In summary, existing research works and systems

only provide monitoring and tracking features, or

they provide health-related interventions, but only fo-

cus on single aspects such as posture. None of the re-

search works contain information on how users per-

ceive such a system during office work. Hence, we

tested a set of health-promoting interventions in a

real-world pilot study with subsequent interviews to

gain information about user perceptions. Being aware

of the users’ valuation of interventions is of vital im-

portance to ensure the further successful development

of health-promoting intervention systems, since inter-

ventions that are not rated as pleasant and helpful by

the user will not be used on a regular basis.

3 CareCam FEATURES

The interventions for our case study were imple-

mented into the existing software CareCam (Kraft et

al., 2021). CareCam enables the objective measure-

ment of important vital data at the workplace using a

simple webcam. This data comprises:

• Pulse rate and pulse rate variability

• Blink frequency

• Upper body posture

• Human emotion through facial expression

recognition

Some representations of the measurements are

displayed in real-time on a user interface. Although

the interface was only relevant in our study to start

and stop the software, the CareCam dashboard is

shown in

Figure

1 to illustrate measurements. The top row

shows the pulse rate in beats per minute and the pulse

rate variability in milliseconds. Below that, a rating

of the upper body posture as either “good” or “bad”

is presented. Next to that, the distance of the eyes

from the screen is displayed in centimeters. The bot-

tom row shows the total number of blinks counted up.

The proportion of different emotions is shown di-

rectly within the displayed camera image. We further

detail the health interventions offered in the following

section. All data of the system was stored locally. No

data was sent to the outside, and no images or video

streams were stored.

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

568

Figure 1: CareCam Dashboard.

4 DESIGN OF INTERVENTIONS

AT COMPUTER

WORKSTATIONS

To help understand the results of the presented user

perception study, this section describes an extended

version of the technical implementation of the inter-

vention system that has already been described in our

previous work (Kraft et al., 2022). For our study, two

types of interventions have been explored: reminders

and breaks. Several articles (cf., e.g. (C. Chen et al.,

2012; Mary & Munipriya, 2011; Paliyawan et al.,

2014 - 2014a)) already included initial forms of alerts

or reminders as health-promoting methods. Hence,

several instances of this intervention type were in-

cluded, which are described in Section 4.1. In addi-

tion, (Herrera et al., 2021) have introduced breaks fo-

cusing on posture-related exercises. Since regular

breaks are an important part of a healthy workday,

breaks should be suggested to the participants in our

case study, too. We decided that break time should

preventively counteract not only bad posture, but also

other typical health risks associated with computer

screen work. Breaks as intervention type are de-

scribed in Section 4.2.

In addition, users can activate a so-called meeting

mode to prevent disruptions caused by reminders or

breaks. Given that meetings are part of the working

time, breaks that would have been scheduled during

the meeting are displayed afterwards.

4.1 Reminders

Four different reminders were implemented: blink,

dynamic sitting, distance to screen, and motivation.

Reminders should not interrupt the workflow.

Therefore, they contain only a headline and a de-

scriptive sentence. They appear as a pop-up notifi-

cation via the operating system. In addition, the re-

minders contain different icons so that the messages

have a recognition effect and can thus be grasped

more quickly.

The four reminders displayed by the system are

shown with their triggers in Table 1. The blink re-

minder is intended to support the user in maintaining

a blink frequency high enough to prevent the rupture

of the tear film (Schmidt, 2008), even during concen-

tration phases on the screen. The threshold of 12

blinks per minute was chosen because the normal

spontaneous blink rate is between 12 and 15 blinks

per minute (Doughty, 2001).

The dynamic sitting reminder was implemented,

based on recommendations by (Mohokum & Dördel-

mann, 2018) to change the sitting position frequently.

For this purpose, the distance of the posture points

recorded by the CareCam is used. If the distance

changes by 12 pixels, this is classified as a movement.

The reminder is triggered if there is no

Table 1: Implemented reminders.

System Message Trigger

blink rate < 12 blinks per minute

no major movement within 2 minutes

distance < 50 cm for 30 seconds

predominantly negative facial expression

within one minute

User-Perception of a Webcam-Based Intervention System for Healthy Habits at Computer Workstations

569

movement after 300 captured images, which corre-

sponds to approx. two minutes.

The distance to screen reminder helps keeping the

right distance between the monitor and the eye. This

is important as eye fatigue should be avoided (Mo-

hokum & Dördelmann, 2018) and maintaining dis-

tance affects the posture, too. Depending on the activ-

ity, the distance between the eyes and the screen is

determined based on the screen's size or the charac-

ters' height. The minimum suggested distance (for

small 13-inch screens) of 50 cm has been specified

for the distance reminder for simplicity. Finally, fa-

cial expressions may, e.g., contain clues to stress and

anxiety (Giannakakis et al., 2017).

An attempt to counteract negative emotions is im-

plemented through the motivation reminder. Once per

second, the emotional state is stored. To exclude

short-term reactions that do not have a larger effect,

we chose a time window of one minute as decisive.

After one minute, when 60 emotion states have been

stored, the predominant emotion in terms of fre-

quency is determined. If the facial expression is pre-

dominantly characterized by a negative emotion (sad-

ness or fear), the reminder is triggered. In this pro-

cess, different motivational messages are displayed

randomly. Overall, to prevent a reminder from ap-

pearing too frequently and thus interrupting and dis-

turbing the user during work, each triggered event is

stored in a buffer. Only after five minutes have

passed, this reminder can lead to a notification again

by the trigger.

4.2 Breaks

Breaks are suggested regularly based on the health

data recorded by CareCam. Breaks are divided into

active breaks that are guided by the system and free

breaks. Active breaks are filled with exercises, while

the user can spend free breaks individually. Exercises

do not just focus on muscle contraction as imple-

mented in (Herrera et al., 2021). In contrast, CareCam

can provide breaks with versatile exercise contents

that, beyond posture, address other topics such as

breathing exercises or mental aspects (e.g., mindful-

ness). According to (Mohokum & Dördelmann,

2018), the rule of thumb is to take a screen break of

five to ten minutes every hour. Therefore, it was de-

cided to include as intervention five-minute breaks

time after 55 minutes of screen time. As it can be rec-

ognized whether the user is in front of the camera at

the computer workstation, individual breaks can be

taken into account. Times that the user is absent from

the screen do not count into the 55 minutes. The user

is informed via system notifications when an active

or free break is upcoming, as can be seen in Figure 2.

Figure 2: Break notifications.

Exercise interventions can be divided into three

groups: “eyes”, “posture”, and “mindfulness”. This

division was made because visual problems, muscu-

loskeletal problems, and stress are among the most

common computer-related health problems (Mary &

Munipriya, 2011). To decide which intervention

group has priority during the break, the reminders are

counted during the 55-minute work period and the

highest group counter is prioritized for the suggested

exercise. The categorization can be seen in Table 2.

As blinking frequency and maintaining distance

from the screen both influence the health of the

“eyes”, the counter for this intervention group is in-

creased by one each time one of these reminders is

triggered. Given that a short distance to the screen can

cause the head to bend forward and thus promote the

development of a “turtle neck”, the counter for the

Table 2: Exercises suggested for active breaks.

Group Exercises in the group Events that increase the group counter

Eyes Relaxation of the eyes, change of directions of gaze Display of blink reminder,

displa

of distance reminde

Posture Stretching and loosening exercises, standing breaks Display of dynamic sitting reminder,

displa

of distance reminde

Mindfulness Deep breathing Display of motivation reminder,

ever

two wor

intervals

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

570

“posture” intervention group is also increased by

one. Additionally, the reminders to sit dynamically

increase the counter for this intervention group by

one, as static muscle activity can lead to tension. The

“mindfulness” intervention group is increased by one

by the motivation reminder. Furthermore, the counter

is increased by one every two work phases of 55

minutes each. This was decided because working

long hours can be a factor for stress that could be mit-

igated with mindfulness exercises, which get a higher

priority with the increased counter. The typical ap-

pearance of an exercise is presented in Figure 3. The

exercise contains an image illustrating what to do to-

gether with a short textual description. Furthermore,

a progress bar depending on time is shown.

All exercises are meant to be performed within the

scheduled five-minute break. However, the perfor-

mance of the exercises does not completely fill in five

minutes. This ensures that the break time does not

have to be exceeded. In that way, the user has enough

time to read through the exercise and can furthermore

use the remaining break individually, e.g., to get a

glass of water, to exchange ideas with colleagues or

simply to have an additional mental break without a

task. After an intervention group has been triggered,

the counter for this group is reset. The counters for

the other groups remain the same in order to balance

the occurrence of intervention groups. In addition,

with this implementation a group – and thus a poten-

tial health issue – can be addressed again if it stands

out. If the software is restarted, e.g., on the next work-

day, the counters for all groups are back to zero.

Free breaks can be used however the user wants

to. A change of work activity is also possible. This

notification occurs when all health parameters are un-

eventful: i.e., when no reminder has been triggered,

or the counters for all intervention groups are zero.

Figure 3: Exemplary Exercise.

5 USER-PERCEPTION OF THE

INTERVENTION SYSTEM

The emphasis of this work is to examine how real-

world users perceive different types of camera-based

health interventions. Therefore, we conducted a pilot

study providing the running system. The approach of

our case study is described in Section 5.1. Through

the test phase with subsequent interviews, infor-

mation was collected on the perception and design of

the health interventions as well as on further func-

tional wishes and aspects, which are presented in Sec-

tion 5.2.

5.1 Case Study Approach

The study was planned as an initial exploration with

a convenience sample. Five research assistants and

two students registered for the study and were pro-

vided with the software to test it during one work

week. Our case criterion is knowledge-intensive work

at the computer workstation. The participants come

from the field of computer science, and thus, a certain

IT affinity can be assumed that might be helpful in a

first software test. The average age of the participants

was 31 years. Five women and two men participated.

The software was tested in the everyday work envi-

ronment by the participants. In advance, the partici-

pants received instructions via a PDF file so that they

were able to install the software on their preferred de-

vice. They had the control to start and stop the soft-

ware, and thus, they had control over the webcam.

Furthermore, measurements and triggered interven-

tions were only stored locally. After the test phase,

semi-structured interviews were conducted in Ger-

man. For this article, the participants’ statements were

thus translated. The interview answers were evaluated

with the help of a inductive categorization method

similar to (Mayring & Fenzl, 2014). The categories

are oriented towards the following objectives and top-

ics of the interview guide:

• Perception of the reminders and breaks

• Functionality of the meeting mode

• Acceptance of the software

• Desired features

5.2 Findings

5.2.1 Perception of Reminders

Regarding the general perception, the icons were

mainly perceived as positive and helpful for faster

recognition of the reminders. It was thus “not such a

User-Perception of a Webcam-Based Intervention System for Healthy Habits at Computer Workstations

571

big distraction”. The different colors of the icons were

emphasized by three people as a recognizable feature.

After getting used to the reminders, they “strongly as-

sociated the message with the color.” One participant

was generally less aware of the reminders due to the

layout of her workstation, through which these did not

appear on her main screen. Thus, she felt the program

should optimally offer the possibility to set the screen

on which the reminder is displayed. Two other test

persons also described that due to the structure of

their workstation in home office, they only saw the

reminders “out of the corner of their eye” and thus

mostly overlooked or “barely noticed” them.

The frequency of the reminders was perceived

very differently depending on the reminder type. Ap-

parently, the time of day when work is done, and the

duration of work can also influence the perception of

the frequency. One participant described the fre-

quency as being “just right for the normal work situ-

ation and normal work hours.” However, when ex-

haustion from work becomes noticeable, “more indul-

gent reminders” were longed for by “increasing the

time interval a bit.” Due to habituation, one person’s

reminders “faded into the background”. Nevertheless,

there was also the case that the reminders appeared

very frequently at the beginning and became less fre-

quent due to compliance. The effectiveness of the re-

minders was, therefore, different for the participants.

The blink reminder was triggered very fre-

quently for most participants. Three participants re-

ceived the blink reminder very frequently, “all the

time”, rating the blink count as inaccurate in conse-

quence. On the contrary, two other participants con-

sidered the blink reminder to be “very helpful”, with

one person also receiving the reminder very fre-

quently. In this case, the blink count was not doubted.

The other person rated the blink reminder as “totally

good” because the usefulness behind it was familiar.

Therefore, for this person, “it actually did not matter

whether it really came at the right moment, whether it

was plausible, but it was just good to be nudged

again.” The remaining two participants could not re-

ceive blink reminders or did not perceive them. It was

also suggested that the unit of blinks per minute

would be useful to support self-reflection.

The frequency of the dynamic sitting reminder

was experienced very individually: the descriptions

ranged from “rather seldom”, “regularly” to “now and

then a bit much”. Four of the six participants who had

noticed the reminder considered this reminder to be

useful and helpful. However, the fit of the trigger tim-

ing was experienced differently. For two participants,

the reminder was not specific enough, as it was not

described in which way the position should be

changed. For this, advanced posture recognition was

desired, for example, to be able to display the tip

“Lean back” in case of a forward-leaning posture.

The distance reminder was rarely displayed for

the most part. Three participants did not receive this

reminder or did not notice it. One person found the

reminder to be good and suitable. Another participant

had problems keeping the distance “since my desk is

not particularly large and that automatically makes

me sit closer.” This means that the distance adjust-

ment could not be made, which resulted in the re-

minder not leading to any change in behavior. In ad-

dition, the idea was expressed once to even combine

distance detection with posture detection so that the

forward head posture, “this slight hump that you

make when you bend over”, is additionally recog-

nized. This participant described the reminder as “the

way it is now, not quite so reasonable”.

The motivation reminder was hardly triggered in

the test phase. Two participants turned off this fea-

ture, because of technical issues or because emotions

were mostly neutral. Of the other people, only one

person received the motivation reminder once and de-

scribed it as follows: “Well, I thought it was funny, it

was kind of cool that it was a quote.” The motivation

reminder was explicitly wished for more once: “to

loosen up quite nice, quite interesting.” However, due

to quality issues with the emotion recognition and/or

often neutral emotions, the trigger for motivation re-

minders should be rethought for future systems.

5.2.2 Perceptions of Breaks

The general perception of breaks and break exer-

cises has been positive, and participants expressed in-

terest in the exercises. However, two participants

could not perform active breaks, because of technical

issues or because they were not noticed or were prob-

ably not displayed, caused by forgetting to deactivate

the meeting mode.

Regarding exercise content, in several inter-

views, the relevance and wish for varied exercises had

been highlighted “to keep it a bit exciting”. Two test

persons emphasized that it was good “if you did not

know the exercises yet”, they received “really always

nicely different” exercises, a sort of “surprise effect”.

Two test persons, who in contrast also received a re-

peated exercise, perceived this negatively and did not

do the exercise that appeared twice. One person justi-

fied this with the type of exercise because she found

it “boring” and rated eye exercises on the screen as

rather unsuitable and mentioned that alternatively, it

would be more helpful to “stand up, look out of the

window, look into the distance”. Eye exercises on the

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

572

screen would have to be “rather something entertain-

ing, if necessary at all”, such as colorful searching im-

ages. For the other person, the frequency of the inter-

ventions was stated as a reason for not doing the re-

peated exercise.

However, the frequency of breaks was rated dif-

ferently by the test persons, with most of them finding

the frequency “exactly right” and “sufficient”. A

break once an hour thus proved to be suitable for the

most part. Only one person emphasized that there

were too many active breaks in general. There is a

preference here for less active breaks, but still receiv-

ing a break reminder every hour. In this short test

phase, only one participant knowingly received a free

break. It would be conceivable, for example, to alter-

nate between a free break and an active break to allow

a little more freedom for the arrangement of the

break. In addition, more acute interventions could be

suggested independently of the time interval if certain

values are strongly abnormal, as requested by one

participant.

The length of the exercises has received different

reviews. Generally, a break time of five minutes was

set, but the exercises did not have to fill the time com-

pletely. Not all participant recognized the correspond-

ing notice in the tool instructions. One user described

that the interventions “went on for what felt like an

eternity”, and half the time would have been enough.

For more advanced interventions, the progress bar

should be tailored for the different exercises and, if

necessary, display the remaining break time after the

exercise is completed to avoid confusion. Other par-

ticipants described the exercises as “all very short” or

easily integrable into their daily work routine, as they

were “always just like five minutes [to take] a quick

break.” Another participant also expressed an idea

that it would be ideal if, during exercise, the software

could “detect and then count down” when an exercise

was being performed in front of the camera. This

could make “the system feel more alive”, and it would

have an “interaction factor.” A type of reward after

successful completion of the exercise would have an

additional motivating effect.

In four interviews, it became apparent that more

information about the reason and goal of a partic-

ular intervention being suggested is a meaningful mo-

tivation component. One user points out that “one or

two sentences” in a “pop-up would be absolutely

enough”. Furthermore, it was emphasized that the fast

and easy comprehensibility of the exercise, “in a few

seconds”, is important. The exercise must be “unam-

biguous in its description” and “described in an un-

derstandable way”. The “ease of execution is [...] a

major criterion” so that a user can “easily integrate

the exercise into the daily work routine.” The length

of the exercise is also an important factor. Here, too,

a balance must be found so that the exercises “have

an effect” but are “not too long”. The current exer-

cises were mostly rated as “well [...] applicable” and

“sufficient”.

The preferred format for the exercises is a video

or animation for six of the seven users. The other par-

ticipant prefers “simply designed things”. She per-

ceived the mindfulness exercise “as totally helpful”,

which includes a very simple animation with little

text. In general, less text description is favored. The

abstraction of the current intervention images could

be kept for videos or animations: One test person

found it “very appealing […] to work with abstract

figures”. Another test person emphasized: “Well, it

does not have to be a human being demonstrating it

to me.” Apart from that, the suggestion to underlay

“mindfulness exercises [...] with music” was ex-

pressed by a user.

5.2.3 Usage of the Meeting Mode

Two of the seven users tried the meeting mode. Once,

the meeting mode was tried at the beginning, but the

software led to difficulties. This is due to the exclu-

sive access of CareCam to the camera so that no other

program could access the camera, while access is also

needed for meetings. As a result, it was completely

turned off for meetings by the participant. As the

other test person had complications with the perfor-

mance of the software, he decided to switch off Car-

eCam for meetings right away. The meeting mode has

been partially forgotten, causing inconvenience to

one user: An intervention occurred in the middle of

an online presentation because the meeting mode was

not activated. Another user “did not re-check every

time” to see if meeting mode had been deactivated

again. As a result, reminders and breaks were also un-

knowingly suppressed during other work activities.

At one point, it was emphasized that in contrast to

breaks, the reminders would be interesting and “even

more useful than [...] under stress”, especially during

video conferences. The reminders “are not so intru-

sive that they distract you from a conversation.”

5.2.4 Software Acceptance

A general interest in the functions of the CareCam is

present in six out of seven participants. Five of them

could imagine using the software permanently during

high screen time. Only one participant would rather

not use the software in everyday working life, be-

cause the person has the impression that it is “quickly

in the background. That it's just an app that runs on

User-Perception of a Webcam-Based Intervention System for Healthy Habits at Computer Workstations

573

the side [...]”. Another person is unsure because, on

the one hand, the CareCam gives her the feeling of

being “watched all the time”, as “the camera is run-

ning all the time”. On the other hand, the software is

perceived as “totally exciting” and “informative” be-

cause it offers many possibilities and draws attention

to health data that are otherwise overlooked. Two of

the six interested individuals did not mention any

concerns with using the software. Using it as a local

tool gave one participant a feeling of security that

eliminated concerns. Concerns arise mainly on data

privacy because it is “something quite personal if you

are constantly being filmed in front of your screen”.

For data processing, “absolute transparency and a

guarantee that the data will not go anywhere” is de-

manded for using the CareCam. It could be helped by

“an agreement” assuring that no data is stored or dis-

tributed. Besides, not every user seems interested in

all of CareCam’s features. Personalization options for

the software could additionally increase acceptance.

5.2.5 Desired Features

The following lists of features were gathered from the

various interviews and could increase user satisfac-

tion as enhancements to our existing system. A

grouped overview is provided in Figure 4. Partici-

pants were asked in the interviews what further types

of and features for reminders they would desire:

Regular breathing: Two participants could imag-

ine a reminder that controls the regularity of breathing

and reminds them to focus on breathing when it is ir-

regular, or breathing interruptions occur.

Advanced posture detection: Posture improve-

ment is a focus of several users. The current reminder

about the sitting position only analyzes whether the

user moves regularly. It lacks a more precise assess-

ment of the sitting position to be able to issue more

accurate reminders for changing the position. Further-

more, it is possible to evaluate the sitting position

based on the quality. In case of an unfavorable pos-

ture, another reminder could intervene. Three partici-

pants have considered using an additional camera

placed on the side to perform more advanced posture

analyses. These analyses could be used to better esti-

mate the sitting position, e.g., detecting a forward-

leaning posture and send more precise reminders.

Display working time: One user has requested an

intermediate output of the working time spent at the

screen workstation.

Individually configurable reminders: Different

users may have individual habits they want to pre-

vent. Simple, individually configurable reminders

could provide a solution. Time-triggered reminders

would be an option for realizing this. For example,

the system could send reminders not relevant to every

user, e.g., for drinking enough water. In addition, it

should be possible to deactivate or adjust the triggers

of the other reminders. If a reminder is perceived as

negative or annoying, this may also affect the percep-

tion of the other reminders. Next, it should be possible

to configure the time interval and, thus, the frequency

between the reminders. For example, different inter-

vals at different times of the day or after long working

hours can be required.

The participants also expressed further wishes and

ideas for active breaks:

Snooze function: It may happen that a break is not

suitable at the moment. A kind of snooze button could

allow the user to “finish what he is working on” so

that one’s thought process is not interrupted. How-

ever, a snooze function should be integrated with care

to prevent the user from always skipping the interven-

tions. One way of implementing this could be to allow

the snooze function to be used only once for each

break, e.g., the user could get “10 minutes, but then”

the break should be taken.

User preference: The user should be allowed to

define his or her priorities. A kind of “user profile” so

that the program can “provide even more individual-

ized tips” to improve the achievement of personal

goals and can more frequently suggest “exercises or

interventions that match these”. The specification of

priorities could also be done after a “kind of initial

phase” in which different exercises are shown. Rating

options could additionally be used so that the system

can learn what kind of exercises are liked by the user.

The frequency of the exercises and the preferred

length of the breaks should also be specified as setting

options, e.g., to opt for more frequent shorter breaks.

Interactivity: The progress bar for breaks should

be tailored for the different exercises. Automatic con-

trol of exercise performance could even make time-

dependent bars obsolete and make the system feel

more interactive. After completing an exercise, a re-

ward system could increase motivation.

For motivation and a better understanding of the

software, the participants expressed the necessity to

include extra information. Desired features in that

direction are:

Setup assistant: The first start of the software

could be supported by a setup assistant. Among other

things, the correct positioning of the camera and the

various setting options should be shown. In general,

an explanation that introduces the various parameters

of the CareCam would increase understanding of the

software. Detailed, general explanations of health as-

pects are also desirable, e.g., the optimal posture at

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

574

the workplace or the correct distance to the screen.

Hints as to why certain things should be maintained

could contribute to increasing motivation.

Info button: The explanations and information

presented by the setup assistant should be “always ac-

cessible”. At times in between, “not the full explana-

tion program” will be appropriate. Using info buttons

could maintain a manageable software layout while

providing additional information. This allows the in-

formation to be re-read as needed, and the user is not

“overwhelmed with information”. In this way, further

information about the reasons a reminder or interven-

tion appears and additional “more specific tips” could

be integrated. Also, information about data privacy

should be available at all times.

Concerning the noticeability of CareCam inter-

ventions, the following features were mentioned:

Sound support: A customized sound would be

useful to differentiate CareCam notifications from

other notifications. Overlooking reminders and

pauses could probably be reduced with this.

Display setting options: It should be possible to

select the screen on which the notifications are dis-

played, as well as the duration of the display to further

reduce overlooking.

Meeting mode: Enabling or disabling the meeting

mode had been forgotten in some cases. Although not

explicitly stated by the user, automatic detection of

whether the user is currently participating in a meet-

ing could solve this problem. Finally, activation of re-

minders separate from breaks should be possible, e.g.,

to still receive reminders during meetings.

Figure 4: Overview of further features.

User-Perception of a Webcam-Based Intervention System for Healthy Habits at Computer Workstations

575

6 DISCUSSION, LIMITATIONS

AND FUTURE WORK

6.1 Discussion of the Findings

To summarize our research, we conducted a first real-

world user test of a prototype system for webcam-

based health-promoting interventions derived from

health literature. The key characteristic of the system

is that it seeks to nudge the user towards healthy

working habits via interventions in the form of re-

minders and (active) breaks. Both are initiated proac-

tively by the system and are sensitive to the users’

current work behaviors and emotional states. This is

made possible by permanently running computerized

analyses of the user in front of the workstation lever-

aging advanced image processing capabilities for au-

tomated recognition of blink, posture, distance, and

emotion. This recognition of the current users’ state

is complemented by a system of counters. According

to the counters, it is periodically decided which is the

next best health promoting intervention for the indi-

vidual user.

Since the user perspective of such interventions is

a still an under-researched topic, we emphasize the

empirical analysis of the users’ perceptions of our

prototype system in the work at hand. To do so, we

conducted a first case study in which seven partici-

pants used the system for approx. one week in their

real-world working environment during their daily

work. Subsequently, we conducted and analyzed in-

terviews which has led to important insights about the

functioning, usefulness and ease of use, and overall

acceptance of such a system. Moreover, we did also

identify improvement potentials for the prototype

system. In the following, we briefly discuss our main

results.

In general, reminders were perceived as useful,

whereby clearly dynamic sitting and blink reminders

received more positive comments than distance and

motivation reminders. This can be explained largely

by implementation deficiencies of our system, where

distance reminders and motivational messages did not

work accurately for all users and did not show up for

all of them. The general positive perception of our re-

minders is consistent with reminders being the num-

ber one feature that is implemented in 72% of the be-

havior change systems according to a recent study

(Villalobos-Zúñiga & Cherubini, 2020). Addition-

ally, since the reminders of our system depend on the

observed current user behavior and emotional state,

they thus adapt to the user and in this sense are per-

sonalized. This may contribute to and explain the

overall positive feedback on the reminders and would

be in line with empirical findings that attest personal-

ized reminders a much higher acceptance than uni-

form reminders (Alhasani & Orji, 2022). Further, the

effectiveness of reminders varied across different

subjects since some subjects observed a decrease of

reminders due to their beginning compliance, i.e.,

habit change, while others did not notice that. The

person-specific effectiveness is also emphasized by

large empirical studies in the field of persuasive sys-

tems that call for personalization as a major design

principle to raise effectiveness (Yfantidou et al.,

2022). Future versions of our prototype system could

adapt even more to individual user preferences, e.g.,

by providing more fine-grained control over reminder

settings. Also, situation awareness could be increased

by automatically detecting meeting situations based

on computer usage or access to the users’ calendar.

Beyond personalization, the general motivation of

users is emphasized as a driver for system use in the

context of self-tracking (Feng et al., 2021). In regard

to deeper underlying motives, also coined as “gener-

ative mechanisms”, it is known that amongst others,

the longing for self-improvement, confirmation and

self-knowledge (Rieder et al., 2021) drives system

use. It can be assumed that all test persons in our sam-

ple implicitly were driven by such motives, so results

might differ for other audiences.

Further, breaks were appreciated, although pref-

erences for breaks differed considerably across us-

ers. For some users, having a five-minute break per

hour was just right, while others preferred longer

working periods or shorter breaks. Once more, this

calls for advanced personalization, which is in line

with literature on persuasive systems (Yfantidou et

al., 2022). Regarding the exercises during an active

break, there seems to be a consensus that exercises

should be “a bit exciting”, i.e., not repetitive or bor-

ing, and that video or abstract illustrations are pre-

ferred over textual descriptions. The importance of

providing varied content to users to avoid boredom is

well-known in persuasive systems literature (e.g., see

(Wiese et al., 2020)).

Regarding meeting mode, our main insight is that

users do not switch off reminders completely but

would prefer to still receive reminders that can be fol-

lowed with low cognitive effort, such as dynamic sit-

ting and blink reminders.

Concerning the overall acceptance of the sys-

tem, the CareCam system was perceived positively as

an innovative new tool for more health-aware work

habits. Six out of seven test users had a generally pos-

itive attitude towards the system after they used it, of

which five expressed interest in continued usage. A

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

576

possible explanation for this might be our conven-

ience sampling strategy leading to test users with ra-

ther high interest and motivation. However, since the

system directly addresses health problems such as

back pain via reminders and exercises being highly

plausible countermeasures, another explanation for

acceptance could be deduced from the Health Belief

Model (HBM). In a nutshell, the model posits that ac-

ceptance emerges when serious health issues are per-

ceived and countermeasures are believed to be effec-

tive in mitigating them. This model has also has been

integrated with the rather “traditional” TAM model

for technology acceptance (Ahadzadeh et al., 2015)

as well as with the more recent UTAUT model (Alpar

& Driebe, 2021) showing the significance of the

HBM in terms of acceptance. This amalgamation of

models could provide a valuable theoretical frame-

work for future research on acceptance. Further, a rel-

evant issue concerning acceptance was data privacy.

Although all data was stored locally and nothing was

transferred, some users expressed concerns about po-

tential privacy issues. Clearly, this calls for further re-

search on how to mitigate potential trust and privacy

issues which could be informed by the Privacy Cal-

culus as theoretical lens to study data sharing willing-

ness for self-tracking data (Dincelli & Zhou, 2017),

also in the work context (Toftgård, 2022). However,

acceptance will considerably depend largely on the

usage scenario. For example, if the system is provided

by a trusted third party such as a public health insur-

ance company, acceptance might differ in contrast to

if the employer provides it.

For further development, the largest group of im-

provement suggestions, i.e., more than one-third, re-

late to advanced personalization options (cf. Fig. 4).

This again is in line with the extant recent literature

in persuasive systems that emphasize the personaliza-

tion as a basic requirement (e.g. (Cho et al., 2022;

Coşkun & Karahanoğlu, 2022; Yfantidou et al.,

2022)). Similarly, it is reflected by the research

stream on self-tracking technology abandon behavior

that, amongst other factors, identify misalignment be-

tween personal goals and the capabilities of technol-

ogy as decisive for discontinuance (Lazar et al.,

2015). Personalization also offers the possibility to

find a better balance between health goals and the sit-

uational requirements during work. For example, in a

nightly work session shortly before a deadline, a user

might want to receive less active break reminders or

even switch them off, whereas, on the following day,

they are highly appreciated again. However, in-

creased self-responsibility comes at the risk that users

apply settings that undermine the effectiveness of the

self-tracking tools. To mitigate this, default settings

grounded in health research in conjunction with warn-

ings when the user applies extreme settings could be

used. Finally, health is an individual topic that affects

everyone, but different aspects can be the focus. With

advanced personalization options, the self-responsi-

bility of users is emphasized. Configuring one’s own

personal intervention system could increase the will-

ingness to execute exercises and comply with self-ad-

justed interventions (i.e., be consistent with and stick

to one’s own goals). This argument is closely con-

nected to research on how technology becomes a de-

vice for augmenting one’s willpower (volition) that is

required to actually execute actions corresponding to

one’s motives (Hamari et al., 2014; Schroeder et al.,

2021). The distinction between motivation and voli-

tion is also an integral part of the Health Action Pro-

cess Approach (HAPA) (R. Schwarzer, 1992). In this

direction, the CareCam prototype supports both, the

motivational phase where the user applies personal

settings based on his or her motivation, and the voli-

tional phase, where the user is nudged to execute the

own-desired behavior. Beyond personalization, some

improvement suggestions of test users related to the

types of reminders, the contents and interaction fea-

tures of active break exercises, the information that

the system provides, and the noticeability of remind-

ers (cf. Fig. 4).

6.2 Limitations of Our Research

The most obvious limitation of our first and prelimi-

nary evaluation with test users is the rather small sam-

ple size. Nevertheless, with real users in a real envi-

ronment doing their ordinary work our evaluation

qualifies as an early ex-post naturalistic evaluation

according to the evaluation strategy selection frame-

work from Pres-Heje and Baskerville (2012) (Vena-

ble et al., 2012) or an EVAL3 (proof of applicability)

or even EVAL4 (proof of usefulness) according to the

framework of (Sonnenberg & vom Brocke, 2012).

However, we still consider our evaluation as early and

preliminary, i.e., as a formative evaluation, where

rich user feedback is collected to further improve the

artifact. Future, more summative evaluations with a

larger sample size allowing for quantitative analysis

with more standardized question items, e.g., from the

Unified Theory of Acceptance and Use of Technol-

ogy (UTAUT 2) model (Venkatesh et al., 2012), are

still do be done. However, early evaluations in the

context of new and innovative artifacts could be con-

sidered analogous to the preclinical phase of develop-

ing a new medicine – the later effect in the living or-

ganism cannot be determined at the stage of develop-

ment of this new substance. Rather, it has to be tested

User-Perception of a Webcam-Based Intervention System for Healthy Habits at Computer Workstations

577

later on in clinical studies (Karagiannis, 2010). An-

other limitation is the period of our evaluation being

one week. In future research, it would be interesting

to run a longitudinal study to analyze more long-term

A, B, C outcomes (Attitudes, Behaviors, Compliance

outcomes (Oinas-Kukkonen, 2010) of the system and

study possible habituation effects.

6.3 Future Work and Research

Possibilities

The interviews have shown which developments

would improve the user perception of the system and

thus ensure the success of the health interventions.

The different perceptions and wishes of the par-

ticipants highlighted the need for a higher level of

personalization of the software. Not every user is in-

terested in all features provided by CareCam. The in-

tention to change one's behavior is a central compo-

nent of the effectiveness of health interventions. A

certain degree of self-responsibility through person-

alization options enables users to define goals for

themselves that they really want to reach. The option

to adjust reminders or create own reminders could re-

alize this individual goal setting.

Moreover, the categories of the exercises should

also be prioritized by each specific user. For example,

some users may be interested in improving posture

but do not care about mindfulness. A prioritization

could be inquired at the beginning, which should be

editable at any time. In addition, after each exercise,

the system could request a basic rating of how the user

liked the exercise, allowing the system to adapt the

predefined user profiles as necessary.

The triggers of the current reminders can be fur-

ther refined. Future reminders or interventions, espe-

cially in the area of stress and well-being, could ben-

efit from including measures of pulse rate and pulse

rate variability.

A calibration phase that captures various factors,

such as age, could provide the opportunity to include

these more complex measurements.

Since the motivational messages were only no-

ticed once, this trigger should also be changed. The

reason for this is the dependence on emotion recogni-

tion. Here, further research is needed to examine

whether the intensity of emotions during office work

is high enough. During our test, some users described

that emotion recognition mostly displayed a neutral

state. Besides emotion recognition, the measurements

should also be improved regarding blink detection.

Generally, it is crucial to prevent the reminders

and pauses from being overlooked. The display of the

reminders should be given setting options for this as

well as an individual sound and a design that distin-

guishes itself even more from other messages.

7 CONCLUSION

Many employees spend large parts of their working

time sitting in front of a computer workstation, poten-

tially leading to serious health problems in the long

run. To counteract this, a few monitoring and inter-

vention systems have been developed. However, up

to now, the user perspective is still under-researched.

Hence, we address this gap in presenting results from

a first investigation how users perceive such an inter-

vention system. The system is based on CareCam and

generates health-promoting interventions for blink

frequency, body posture, screen proximity and emo-

tion-based messages. Several subjects used the sys-

tem in a real-world case study during their ordinary

daily work for one week. We then gathered rich qual-

itative feedback based on interviews from which we

distilled insights on how users perceive such a sys-

tem. From a high-level perspective, key results are:

(1) Reminders are an effective instrument to

raise awareness for unhealthy work behaviors and

promote healthy ones. All test persons welcomed the

reminders; surprisingly, no one felt interrupted or dis-

turbed by them.

(2) Active breaks with versatile content are

welcomed, but customization is needed. All users did

express their interest in exercises that must be easy to

understand and perform. The format should include

minimal text and preferably show a video or anima-

tion. To further increase awareness and motivation,

more explanations and information in the software are

needed and have been asked for by most participants.

Further, customization regarding frequency and dura-

tion is needed.

(3) Personalization options are of vital im-

portance. The largest parts of improvement sugges-

tions and feature requests are related to personaliza-

tion. The possibility of fine-tuning the system can in-

crease the effectiveness of reminders and compliance

behaviors for exercises. It moreover emphasizes the

self-responsibility of the user.

Our final conclusion is that the ultimate quest in

designing an effective health-promoting intervention

system seems to be finding a balance between effec-

tiveness and user satisfaction. The user must perceive

the health interventions as comfortable and applicable

as possible to be willing to implement them on a reg-

ular basis. At the same time, interventions must be ef-

fective regarding health-related outcomes.

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

578

All in all, our work contributes to the field of

health-promoting intervention systems for computer-

based work. Our preliminary empirical observations

and findings could be a starting point to inform the

design of future intervention systems.

REFERENCES

Ahadzadeh, A. S., Pahlevan Sharif, S., Ong, F. S., &

Khong, K. W. (2015). Integrating health belief model

and technology acceptance model: An investigation of

health-related internet use. In J Med Internet Res. 2015

Feb 19;17(2), e45.

Alhasani, M., & Orji, R. (2022). SortOut: Persuasive Stress

Management Mobile Application for Higher Education

Students. In N. Baghaei, J. Vassileva, R. Ali, & K.

Oyibo (Eds.), PERSUASIVE 2022 (Vol. 13213, pp. 16–

27). Springer.

Alpar, P., & Driebe, T. (2021). Patients’ Attitudes Toward

Apps for Management of a Chronic Disease. In S.

Stieglitz, R. Schütte, & F. Ahlemann (Eds Information

Systems and Organisation. Innovation Through Infor-

mation Systems: WI 2021. vol. 46, pp. 22–37). Springer.

Rieder, A, Lehrer, C., Eseryel, Y., & Jung, R. (2021). The

Generative Mechanisms Behind Technology-enabled

Health Behavior Change. In ECIS 2021.

Bitkom. (2018). Bitkom Digital Office Index 2018: Eine

Studie zur Digitalisierung von Büro- und

Verwaltungsprozessen in deutschen Unternehmen.

Burkhart, S., & Hanser, F. (2018). Einfluss globaler

Megatrends auf das digitale Betriebliche Gesundheits-

management. In D. Matusiewicz & L. Kaiser (Eds.),

Digitales Betriebliches Gesundheitsmanagement:

Theorie und Praxis (pp. 37–55). Springer.

C. Chen, T. Määttä, Kevin Bing-Yung Wong, & H.

Aghajan (2012). A collaborative framework for ergo-

nomic feedback using smart cameras. In ICDSC 2012.

Cho, J., Xu, T., Zimmermann-Niefield, A., & Voida, S.

(2022). Reflection in Theory and Reflection in Prac-

tice: An Exploration of the Gaps in Reflection Support

among Personal Informatics Apps. In S. Barbosa, C.

Lampe, C. Appert, D. A. Shamma, S. Drucker, J. Wil-

liamson, K. Yatani (Eds.), Proc. CHI 2022 (pp. 1–23).

Coşkun, A., & Karahanoğlu, A. (2022). Data Sensemaking

in Self-Tracking: Towards a New Generation of Self-

Tracking Tools. International Journal of Human–Com-

puter Interaction, 1–22.

Dincelli, E., & Zhou, X. (2017). Examining self-disclosure

on wearable devices: The roles of benefit structure and

privacy calculus. In AMCIS 2017 Proc.

Doughty, M. J. (2001). Consideration of three types of

spontaneous eyeblink activity in normal humans: Dur-

ing reading and video display terminal use, in primary

gaze, and while in conversation. In Optom Vis Sci. 2001

Oct; 78(10), 712–725.

Feng, S., Mäntymäki, M., Dhir, A., & Salmela, H. (2021).

How Self-tracking and the Quantified Self Promote

Health and Well-being: Systematic Review. In J Med

Internet Res. 2021 23(9), e25171.

Fichter, K., Clausen, J., & Hintemann, R. (2012). Roadmap:

»Resource-efficient workplace computer solutions

2020«: Development of a lead market for green office

computing. BMU, Federal Environment Agency &

BITKOM, Berlin, Dessau, Roßlau, Leitfaden.

Giannakakis, G., Pediaditis, M., Manousos, D., Kazantzaki,

E., Chiarugi, F., Simos, P. G., Marias, K., & Tsiknakis,

M. (2017). Stress and anxiety detection using facial

cues from videos. In

Biomedical Signal Processing and

Control, 31, 89–101.

Hamari, J., Koivisto, J., & Pakkanen, T. (2014). Do Per-

suasive Technologies Persuade? - A Review of Empir-

ical Studies. In PERSUASIVE 2014, (vol. 8462, pp.

118–136). Springer.

Herrera, F., Niño, R., Montenegro-Marín, C. E., Gaona-

García, P. A., Mendívil, I. S. M. de, & Crespo, R. G.

(2021). Computational method for monitoring pauses

exercises in office workers through a vision model. In J

Ambient Intell Human Comput 12, 3389–3397.

Karagiannis, D. (2010). Welche Rolle kann bzw. soll die IT

bei der Umsetzung und Unterstützung gestaltungs-

orientierter WI-Forschung spielen? In H. Österle (Ed.),

In Gestaltungsorientierte Wirtschaftsinformatik: Ein

Plädoyer für Rigor und Relevanz. Infowerk.

Kraft, D., Schmidt, A., Büttner, L., Oschinsky, F. M.,

Lambusch, F., van Laerhoven, K., Bieber, G., &

Fellmann, M. (2022). CareCam: Towards user-tailored

Interventions at the Workplace using a Webcam. In

ICPS, PETRA 2022 (pp. 494–499). ACM.

Kraft, D., van Laerhofen, K., & Bieber, G. (2021). Car-

ecam: Concept of a new tool for Corporate Health Man-

agement. In ACM Digital Library, The 14th PErvasive

Technologies Related to Assistive Environments Con-

ference (pp. 585–593). ACM.

Lazar, A., Koehler, C., Tanenbaum, J., & Nguyen, D. H.

(2015). Why we use and abandon smart devices. In K.

Mase (Ed.), Ubicomp 2015 Proc. (pp. 635–646). ACM.

Maeda, N., Hirabe, Y., Arakawa, Y., & Yasumoto, K.

(2016). COSMS: Unconscious stress monitoring sys-

tem for office worker. In P. Lukowicz, A. Krüger, A.

Bulling, Y.-K. Lim, & S. N. Patel (Eds.), Ubicomp 2016

Proc. (pp. 329–332). ACM.

Mary, G. M. J., & Munipriya, P. (2011). Health monitor-

ing of IT industry people. In ICECT 2011 (pp. 144–

148). IEEE.

Mayring, P., & Fenzl, T. (2014). Qualitative Inhaltsana-

lyse. In N. Baur & J. Blasius (Eds.), In Handbuch Me-

thoden der empirischen Sozialforschung (vol. 3, pp.

543–556). Springer.

Mohokum, M., & Dördelmann, J. (2018). Ergonomie am

Arbeitsplatz mit Praxisbeispielen. In M. Mohokum & J.

Dördelmann (Eds.), In Betriebliche Gesundheitsför-

derung (pp. 175–213). Springer.

Oinas-Kukkonen, H. (2010). Behavior Change Support

Systems: A Research Model and Agenda. In T. Ploug,

P. Hasle, H. Oinas-Kukkonen (eds) PERSUASIVE 2010

(vol. 6137, pp. 4–14). Springer.

User-Perception of a Webcam-Based Intervention System for Healthy Habits at Computer Workstations

579

Paliyawan, P., Nukoolkit, C., & Mongkolnam, P. (2014,

May). Prolonged sitting detection for office workers

syndrome prevention using kinect. In ECTI-CON 2014,

(pp. 1–6). IEEE.

Paliyawan, P., Nukoolkit, C., & Mongkolnam, P. (2014,

October). Office workers syndrome monitoring using

kinect. In APCC2014, (pp. 58–63). IEEE.

Park, J. H., Moon, J. H., Kim, H. J., Kong, M. H., & Oh, Y.

H. (2020). Sedentary Lifestyle: Overview of Updated

Evidence of Potential Health Risks. In Korean J Dam

Med 2020, 41(6), 365–373.

R. Schwarzer (1992). Self-efficacy in the adoption and

maintenance of health behaviors: Theoretical ap-

proaches and a new model. In R. Schwarzer (Ed.), Self-

efficacy: Thought control of action, (pp. 217-243)

Schmidt, D. (2008). Tipps und Tricks für den Augenarzt:

Problemlösungen von A bis Z. Tipps und Tricks.

Springer.

Schroeder, T., Haug, M., & Gewald, H. (2021). The differ-

ence between motivation and volition matters! A quali-

tative study on mobile health application adoption. In

ECIS 2021 Research Papers, (pp.1-18).

Sonnenberg, C., & vom Brocke, J. (2012). Evaluations in

the Science of the Artificial – Reconsidering the Build-

Evaluate Pattern in Design Science Research. In K.

Pfeffers, M. Rothenberger, B. Kuechler (eds), DES-

RIST 2012 (vol. 7286, pp. 381–397). Springer.

Taieb-Maimon, M., Cwikel, J., Shapira, B., & Orenstein, I.

(2012). The effectiveness of a training method using

self-modeling webcam photos for reducing mus-culo-

skeletal risk among office workers using comput-ers. In

Applied Ergonomics 2012, 43(2), 376–385.

Toftgård, V. (2022). Employee acceptance from a privacy

perspective of wearable fitness trackers at work: A

qualitative study of employees in Sweden (Dissertation).

Venable, J., Pries-Heje, J., & Baskerville, R. (2012). A

Comprehensive Framework for Evaluation in Design

Science Research. In DESRIST 2012 (vol. 7286 pp.

423–438). Springer.

Venkatesh, V. Thong, J.Y., & Xu, X. (2012). Consumer Ac-

ceptance and Use of Information Technology: Extend-

ing the Unified Theory of Acceptance and Use of Tech-

nology. In MIS Quarterly, 36(1), 157-178.

Vildjiounaite, E., Huotari, V., Kallio, J., Kyllönen, V.,

Mäkelä, S.-M., & Gimel’farb, G. (2018). Detection of

Prolonged Stress in Smart Office. In K. Arai (Ed.), SAI

2018, (vol. 857, pp. 1253–1261). Springer.

Villalobos-Zúñiga, G., & Cherubini, M. (2020). Apps That

Motivate: a Taxonomy of App Features Based on Self-

Determination Theory. In Int J Hum Comput Stud, 140,

102449.

Wiese, L., Pohlmeyer, A. E., & Hekkert, P. (2020). Design

for Sustained Wellbeing through Positive Activities—

A Multi-Stage Framework. In Multimodal Technol. In-

teract. 4(4), 71.

Yfantidou, S., Sermpezis, P., & Vakali, A. (2022). 12 Years

of Self-tracking for Promoting Physical Activity from a

User Diversity Perspective: Taking Stock & Thinking

Ahead. In A. Bellogin, L. Boratto, O. C. Santos, L. Ar-

dissono, & B. Knijnenburg (Eds.), UMAP 2022 (pp.

211–221).

Scale-IT-up 2023 - Workshop on Best Practices for Scaling-Up Digital Innovations in Healthcare

580