Heart Rate Visualizations on a Virtual Smartwatch to Monitor Physical

Activity Intensity

Fairouz Grioui

and Tanja Blascheck

Institute for Visualization and Interactive Systems, University of Stuttgart, Germany

Keywords:

Micro Visualization, Virtual Smartwatch, Heart Rate Visualization, Fitness Data, Empirical Study, Virtual

Reality.

Abstract:

We investigate three visualizations showing heart rate (HR) and HR zones (HRZ) data collected over time

and displayed on a virtual smartwatch, to monitor physical activity intensity. To understand exercise behavior,

we ﬁrst conducted a survey with 57 participants and found that most of them track their activities (66%)

using wrist wearable devices (i. e., smartwatches or ﬁtness bands) and that during the course of the exercise

data is primarily represented as text or a combination of text and icon. To support reaching a speciﬁc ﬁtness

goal, we designed a bar chart visualization combining both current and historical HR and HRZ data. Among

the three visualizations, two present an additional chart (i. e., a horizontal and radial bar chart summary),

showing the amount of time spent per HRZ (i. e., low, moderate, and high intensity). In a controlled study

performed in virtual reality, we compared participants’ performance with each visualization asking participants

to make a quick and accurate decision while exercising (i. e., playing a tennis-like game). Results from the

study show evidence of a difference in task performance between visualizations with and without a summary

chart—visualizations showing a summary chart performed better than the version without. Finally, based on

our study results we present lessons learned.

1 INTRODUCTION

Wrist wearable devices (e. g., smartwatches and ﬁtness

bands) are now more affordable and people increas-

ingly use them to track their ﬁtness data. According to

a survey conducted with 4,272 American adults (Pew

Research Center, 2022), 21% of them wore a ﬁtness

tracker on a regular basis. By supporting automated

personal data collection and providing their wearers

with instantaneous feedback on their progress, these

modern ﬁtness trackers foster data exploration on the

go; giving the opportunity to real-time physical perfor-

mance regulation to meet a pre-set ﬁtness goal.

When people look at the data during the course

of an activity, usually they consume the information

in short time-lapses—through individual glances of

5 seconds or less (Blascheck et al., 2021). Due to

this consumption of the data as a secondary task, the

presented information has to be informative and al-

low reﬂection in a short time frame. Visualizing the

data can help to convey this information; however, one

of the challenges of designing visualizations for such

a usage scenario is that the provided data should be

https://orcid.org/0000-0002-3927-4821

https://orcid.org/0000-0003-4002-4499

concise to reduce interruption from the primary activ-

ity (Amini et al., 2017).

People using wrist wearable devices when exer-

cising are interested in learning more about their per-

formance and pushing their limits to reach a ﬁtness

goal (Amini et al., 2017). Therefore, we conducted

a survey to understand which types of data facilitate

people to monitor their ﬁtness data while and after

exercising, how their device represents the data, and

what motivates them to track and analyze their data.

Our survey, but also previous work (Amini et al.,

2017; Islam et al., 2020; Neshati et al., 2019a), shows

that heart rate (HR) is one of the most common types

of data people collect with wrist wearable devices. In

the context of ﬁtness data tracking, monitoring HR

data is an efﬁcient indicator to control physical perfor-

mance (Ketcheson et al., 2015; Navarro et al., 2013)

(i. e., to foster exertion or vice versa) and gauge phys-

ical activity intensity (She et al., 2020) (e. g., HIIT

training). This monitoring of HR while performing

an exercise lends itself to designing appropriate HR

visualizations.

To support in-situ health and ﬁtness data explo-

ration on small wearable devices used on the go, we

need to consider different design recommendations

Grioui, F. and Blascheck, T.

Heart Rate Visualizations on a Virtual Smartwatch to Monitor Physical Activity Intensity.

DOI: 10.5220/0011665500003417

In Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2023) - Volume 3: IVAPP, pages

101-114

ISBN: 978-989-758-634-7; ISSN: 2184-4321

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

101

when creating visualizations (Amini et al., 2017; Is-

lam et al., 2020; Neshati et al., 2019a). Therefore,

we leverage the following two considerations in our

visualizations design. First, because the provided rep-

resentations on these small devices often consist of

raw textual information—without contextual historical

data—issues related to misinterpretation of the data

and delayed reactions are susceptible to happen. For

this reason, the use of standard and familiar visual-

ization techniques (e. g., bar, donut, and line charts)

is encouraged to support activities that require reﬂec-

tion (Amini et al., 2017; Neshati et al., 2019a; Choe

et al., 2017). Second, research recommends to com-

bine multiple inter-correlated types of data to easily

estimate the distance to the goal (Amini et al., 2017).

Hence, minimizing the mental load or the necessity to

recourse to external tools. Accordingly, we designed

three HR visualizations using a vertical bar chart (Fig-

ure 6) to show current and historical HR and HR zone

(HRZ) data, as well as a summary of the amount of

time a person spent in each HRZ (i. e., low, moderate,

and high intensity). The latter was 1) encoded using

colors of the vertical bar chart for one of the visual-

izations, and represented as 2) a separate horizontal or

radial bar chart summary for the two other designs.

During the Covid-19 pandemic, it has become

more common to exercise in a virtual environment at

home (Siani and Marley, 2021). In addition, a virtual

environment has ample opportunities to conduct a con-

trolled lab study and to simulate performing a primary

(i. e., exercising) and secondary task (i. e., monitoring

the physical activity intensity), to test the effectiveness

of the three HR visualizations.

Through this research project, we aim to answer

the following research questions:

•

RQ1: Can micro-visualizations on smartwatches

support decision making and reﬂection tasks on

the go?

•

RQ2: How fast and accurately can people explore

complex visualizations (i.e., combining multiple

data types) on small screens as a secondary task?

•

RQ3: How do different representations and data

aggregation techniques inﬂuence data reading and

interpretation?

•

RQ4: To which extent can virtual reality support

simulating real world exercising and a smartwatch

visualization task?

To summarize, our contributions are threefold.

First, we describe our survey’s ﬁndings on ﬁtness data

tracking using wrist wearable devices. Second, we

present three HR visualizations, that we implemented

based on design recommendations from the literature

and existing examples, then empirically assessed their

effectiveness in a self-monitoring task in a VR study.

Finally, we reﬂect on the ﬁndings and our study results

and discuss their implications.

2 RELATED WORK

In the following, we discuss the work related to our

research including micro-visualizations, data explo-

ration in motion, ﬁtness data tracking on wearables

and ﬁtness trackers and health data monitoring in VR.

2.1 Micro-Visualization in Research

Wrist wearable devices are attracting more attention

from different research communities due to their ubiq-

uity in different aspects of our daily activities (e. g.,

monitoring health, ﬁtness or sleep data). While

some research in human-computer interaction ex-

plored novel interaction techniques (e. g., to overcome

the fat ﬁnger problem (Neshati et al., 2021b)), our own

work and a number of previous studies in the domain

of visualization are interested in investigating data

representations adapted to the conﬁguration of these

devices (e. g., small screens (Chen, 2017; Neshati et al.,

2019b, 2021a; Suciu and Larsen, 2018)) and to their

usage context (e. g., visualization in motion (Yao et al.,

2020), glanceable visualization (Blascheck et al., 2019;

Islam et al., 2022)).

Few empirical studies related to visualization have

been conducted to understand representations of micro

visualizations—adjusted data representations visual-

ized on small screens of a few square centimeters.

Blascheck et al. (2019) evaluated participants’ time

threshold using three visualizations (i. e., bar, donut,

and radial bar chart) in a data comparison task. Bar and

donut charts resulted in faster responses. The authors

followed up with a replication study (Blascheck and

Isenberg, 2021) to compare whether the same results

from the smartwatch study could be obtained when

displaying the stimuli on a larger screen (i. e., a lap-

top screen). The authors tested both smartwatch-sized

(320

320 px) and larger visualizations (1280

1280

px) on a laptop screen. The overall results showed no

difference between stimuli sizes. Neshati et al. (2019b)

introduced a novel visualization technique named G-

Sparks—inspired by sparklines (Tufte, 2001)—to in-

vestigate the effect of compression of a line chart (on

the x-, y-, or xy-axis) on task performance. Accord-

ing to the completion time, accuracy and interaction

count, results were in favor of the x-axis compression

technique. Islam et al. (2022) contributed three studies

about sleep data visualizations (i. e., ﬂoating bar chart

and hypnogram) when displayed with three different

IVAPP 2023 - 14th International Conference on Information Visualization Theory and Applications

102

form factors (i. e., Square, Wide, and Tall) simulating

different small devices (i. e., smartwatch and ﬁtness

band). Results showed that the suitable form factor de-

pends on the task. In general, both tested visualizations

were glanceable (i. e., readable in

5 s). Following

these previous works, we contribute a study in the

realm of micro visualizations, with focus on ﬁtness

data exploration.

2.2 Data Exploration in Motion

Despite their omnipresence in the activities of their

wearers (e. g., working, training, cooking), most of

the previous studies involving visualizations on small

devices (Blascheck et al., 2019; Islam et al., 2022;

Neshati et al., 2019b, 2021a) were conducted in a sta-

tionary pose. For example, Blascheck et al. (2019)

ﬁxed a smartwatch on an adjustable stand simulat-

ing the viewing angle and distance at which a smart-

watch is usually held while reading information. In

contrast, the number of studies that explored visual-

izations on small wearable devices while in motion

is limited. Schiewe et al. (2020) evaluated the per-

formance of two graph-based visualizations showing

two foot strike types (i. e., heel and forefoot strikes)

in comparison to textual feedback of the current fore-

foot strike rate on a smartwatch while walking on a

treadmill. They found that the real-time graph-based

visualizations were preferred over textual data repre-

sentations. Similarly, Neshati et al. (2021b) explored

two interaction techniques with graphs on a smart-

watch while walking on a treadmill and when having

a standing posture. Yao et al. (2020) deﬁned a ﬁrst de-

sign space of visualizations used while in motion. The

dimensions of the design space include the entity in

motion, characteristics in motion, as well as the motion

relationship between viewer and visualization. In our

study, we investigate ﬁtness data exploration—as a sec-

ondary task—while playing a tennis-like game in VR.

The player is in a high active mode—displacement of

the arm to the right and left side to catch balls—before

we display the visualization. When the participant

is observing the data, they are in a standing posture

and a minimum degree of movement is expected. The

motion in our case concerns mainly the arm twist and

elevation when looking at the (virtual) smartwatch.

2.3 Fitness Data Tracking on Wearables

In their survey, Islam et al. (2020) found that wear-

ers of smartwatches are mostly interested in tracking

their ﬁtness and health data. However, due to the lack

of knowledge regarding the challenges of data con-

sumption on the small screens of smartwatches, as

well as to identify the speciﬁc needs of people dur-

ing and after a physical activity, previous work also

performed user-centered research to understand the

needs of smartwatch wearers. Amini et al. (2017)

conducted interviews with people who tracked their

ﬁtness data on a regular basis and found that partici-

pants have different needs when the data is explored

on the go (i. e., exploring one or multiple values, es-

timating the progress towards a goal, comparing one

or multiple measures with other people’s and stay-

ing motivated). The authors converted these tasks

into data exploration scenarios and invited nine profes-

sional designers to prototype visualizations—adapted

to the small screen of a smartwatch—that would sup-

port in-situ data exploration during a ﬁtness activity.

For the goal-based category, most sketches were us-

ing visualizations (e. g., donut and pie charts or space

ﬁlling shapes such as a single bar or icon) instead of

text. In addition, in relation to personal data collection,

Choe et al. (2017) organized discussion sessions with a

group of Quantiﬁed Selfers to understand the practices

and techniques they employed to collect and explore

outcomes from their personal data. Based on the gath-

ered feedback, they generalized the reported problems

into three pitfalls (i. e., tracking too many things, not

tracking triggers and context,as well as insufﬁcient sci-

entiﬁc rigor). In an attempt to summarize limitations

related to the exploration and the design of visualiza-

tions on smartwatches, Neshati et al. (2019a) report

a set of challenges related to the hardware (e. g., lack

of memory, limited processing power, limited battery

life) and related to the data representation (e. g., lack

of guidelines for complex data representation on small

screens). They speciﬁed that visualization techniques

should be applied to represented HR data for an in-situ

exploration. These prior studies contributed in build-

ing a better understanding of people’s behavior when

tracking their ﬁtness data. In addition, they informed

a set of general design considerations of in-situ visual-

ization that support reﬂection on the go. We apply two

of the aforementioned design considerations (i. e., pro-

mote visualizations over textual representations and

combine interrelated types of data) to visualize HR

data over time. In addition, we test our three HR visu-

alizations to foster data exploration on the go during

the course of a ﬁtness activity and to convey meaning-

ful information to reach a predeﬁned goal.

HR data is widely used for health and ﬁtness appli-

cations. For example, Albaghli and Anderson (2016)

explored data from the embedded smartwatch sensors

to keep chronic patients aware of their health state and

to facilitate the identiﬁcation of anomalies by means of

visualizations. Also, Muangsrinoon and Boonbrahm

(2017) visualized real-time HR and HRZ data on a

Heart Rate Visualizations on a Virtual Smartwatch to Monitor Physical Activity Intensity

103

Figure 1: Counts regarding the ﬁtness data (i. e., step count, speed, heart rate, elevation, duration, distance, calories, and blood

pressure) that participants reported to check, while exercising (left) and after the exercise (right).

smartwatch, to monitor the physical activity intensity

of a cyclist (on a stationary bicycle). The authors

found that real-time HR monitoring can help exer-

cising within a desired target HRZ. Likewise, in our

study, we explore visualizations of HR and HRZ data

to monitor a ﬁtness activity in a goal oriented task.

2.4 Fitness Trackers and Health Data

Monitoring in VR

HR data was investigated in VR and by means of ex-

ergames to monitor physical activity intensity and to

raise player’s awareness about their exertion rate (e. g.,

to foster exertion (Ketcheson et al., 2015) or to avoid

over exercising (She et al., 2020; Yoo et al., 2018)).

Moreover, wearables are emerging in VR in different

ways. A new concept to explore ﬁtness and game data

in VR is a virtual smartwatch (YUR.Watch, 2020).

The design mimics a real smartwatch with an exten-

sion sleeve. Only information about burned calories

is displayed as text on the watch face. The rest of the

tracked data (e. g., number of squats and current heart

rate) is represented with text and icons and displayed

on the watch sleeve. More recently Queck et al. (2022)

introduced SpiderClip, a system that enables the ex-

ploration of simulated wearables in VR. SpiderClip

consists of hardware and software components, allow-

ing the real-time collection and monitoring of spatial

and physiological data in VR. The authors validated

their design with a proof-of-concept study in which

they tested HR data measurements and visualizations

during a ﬁtness exercise. The authors visualized the

HR data in two different ways: (1) by showing the cur-

rent HR value on the smartwatch display and on an HR

gauge on a separate sleeve, as well as (2) showing the

same data on a heads-up display-like representation.

We also leverage the VR environment to implement

a tennis-like game, in which we can control and as-

sess participants actions while playing the game (e. g.,

count the scored balls) and behavior after visualizing

the data (i. e., measure response time, record answers).

To simulate a real-world experience, we display the

HR visualizations on a virtual smartwatch.

3 SURVEY

We decided to approach people who track their ﬁtness

data through a survey to learn about their tracking

techniques and behaviors while and after exercising to

understand 1) the device they rely on and that support

their activity tracking, 2) the data they frequently col-

lect and how their device presents it, 3) their motives

for self-tracking their exercise behavior, and 4) poten-

tial use of ﬁtness data tracking in VR. We summarize

insights from the gathered data to design a visualiza-

tion based on the needs of people.

3.1 Survey Design

We conducted an anonymous online survey with SoSci

Survey (SosciSurvey, 2022) and advertised it on dif-

ferent social media and professional platforms (e. g.,

Facebook groups, Twitter, LinkedIn, and mailing lists).

We structured the survey into ﬁve sections, which con-

sists primarily of close-ended questions. Our supple-

mental material (OSF, 2022) contains a detailed list of

these questions together with the results of the survey.

The ﬁrst section was concerned with getting con-

sent and ensuring participants were old enough to par-

ticipate. Because we are interested to learn how people

exercise and track their ﬁtness data, an initial question

asked about regular exercise and tracking of data.

Next, we collected general information about the

physical activity and the frequency of data tracking.

We asked participants if they track all their sport ac-

tivities using a ﬁtness tracker, how often they exercise

per week, where they usually practice this activity and

how often they track their individual sport activities.

With the main part of the survey, we intended to

learn about participant’s primary sport activity. Here

we asked what the primary sport activity is, how they

IVAPP 2023 - 14th International Conference on Information Visualization Theory and Applications

104

Figure 2: Number of data collect during an activity.

track their ﬁtness data and when they check their ﬁt-

ness data. We were also interested to learn if there is

a difference in when participants check their ﬁtness

data—while exercising and after the exercise. There-

fore, we asked how the used ﬁtness device represents

this data during and after the exercise. To learn about

reasons why participants track their ﬁtness data, we

also asked them what their primary motivation is to

check their ﬁtness data during and after the exercise.

In case participants tracked more than one physical

activity, we gave them the opportunity to report about

their secondary activity using the same questions as

for the primary activity.

Last, we asked participants if they also exercise in

a VR environment. Those participants who answered

the question with yes were asked when they check

their ﬁtness data when exercising in VR and how their

ﬁtness data is represented. Those participants who do

not track their ﬁtness data in VR, were asked if they

would be interested to do so and how they would like

to represent their ﬁtness data in VR.

3.2 Participants

We recruited in total 57 participants; 16 of them were

excluded at the beginning of the survey: 4 participants

do not practice any sport activity, 11 do not track their

ﬁtness data and one participant was younger than 18

years. The remaining 41 participants conﬁrmed that

they exercise on a regular basis: 24 practice their pri-

mary sport activity up to 2 or 3 times a week, 7 of

them 4 or 5 times a week, 6 once a week, 2 exercise

every day, 1 once every two weeks, and 1 participant

2 or 3 times a week. The majority of the participants

track their ﬁtness data frequently: 27 always track their

physical activity, 13 do it sometimes, and 1 rarely.

3.3 Results

Participants reported a total of 65 sport activities: 41

primary activities and 24 secondary activities. Partici-

pants mostly perform their tracked activities outdoors

(40 of 65), and the most common types of activities

participants perform are running (23) and cycling (17).

3.3.1 Fitness Data Tracking Devices

Participants reported that they track their ﬁtness data

on wearable devices for 43 sport activities: for 31

activities they use a smartwatch and for 12 they use

a ﬁtness band for tracking. In addition, participants

use their smartphone to track 22 activities, 6 activities

were not tracked (e. g., swimming) and one participant

mentioned using a golf-tracking system (i. e., Track-

man Range). This question was designed as a multiple

choice question and 5 participants selected more than

one device for a single activity: 2 used smartwatch and

smartphone, 2 used smartwatch and ﬁtness band and 1

reported to use all three devices.

3.3.2 Data Type and Representation

By comparing the ﬁtness data participants track during

and after their sport activity, we found that duration

(i. e., 56 answers during the activity and 60 after the

activity) and distance (i. e., 56 answers during the ac-

tivity and 56 after the activity) are the top two tracked

types of data (Figure 1). Then come in order: speed,

calories, and heart rate as the data types tracked during

the activity. After the activity, participants reported

that they track calories, step count, and their speed.

Participants tend to track multiple types of ﬁtness

data (i. e., 2 to 8 data types) during the activity (Fig-

ure 2).

Figure 1 reveals that most of the data tracked dur-

ing the course of the physical activity are represented

using either “text” or a combination of “text and icon.”

In contrast, for the data tracked after the activity, par-

ticipants use more “graphs/charts” as well as “graph

and text” representations.

3.3.3 Motivation to Track Fitness Data

Participants rated Checking a ﬁtness goal highly

for both activities (i. e., second reason after “check

progress” with 35 answers while exercising and as

a primary reason with 47 answers after exercising).

Some participants gave examples to explain the ﬁtness

goals they target during the exercise, for example: “I

usually run with my heart rate being between 140 and

155 bpm. This helps to check if I am overdoing things

and if I will be able to ﬁnish what I was planning to

do—running 1h 45?”

3.3.4 Exercising in Virtual Reality

Only 6 participants exercise in VR. Two participants

reported that they do not collect data when exercising

in VR, four check their data after exercising (i. e., 2

on a smartwatch or smartphone and 2 in VR). Asked

if they would like to see their ﬁtness data in VR, 2

Heart Rate Visualizations on a Virtual Smartwatch to Monitor Physical Activity Intensity

105

participants chose to visualize the data on a virtual

smartwatch, the rest of the answers were distributed

based on the different options (i. e., 2 preferred a small

panel to the side, 2 chose to visualize the data on a

larger panel in front of them, 1 chose a small panel in

front of them, and 1 a larger panel to the side).

3.4 Summary

We summarize and categorize the survey’s ﬁndings:

Data Tracking while Exercising. Results show that

for 66.1% of the reported activities, ﬁtness data was

tracked on wrist wearable devices. Despite their re-

stricted display area, participants mostly track 6 differ-

ent types of ﬁtness data during the course of an activity.

This is in line with previous research on watch face

design (Islam et al., 2020).

Data Representation while Exercising. While ex-

ercising (Figure 1, left), participants represent their

ﬁtness data primarily as “text” and “text and icon” com-

binations, not using charts or graphs. The results also

show that participants’ ultimate motivation for tracking

their data is to monitor their ﬁtness progress or goal.

This is in accordance to previous research (Amini et al.,

2017) and could stem from smartwatch manufactur-

ers not providing such charts and graphs, even though

visualization research has shown that participants are

able to read charts and graphs quickly (Blascheck et al.,

2019; Islam et al., 2022; Neshati et al., 2019a).

As a consequence of this survey, in the following

study, we explore visualizations to represent ﬁtness

data (i. e., HR data), then assess their efﬁciency to

monitor a ﬁtness goal while exercising.

4 IN-LAB STUDY

In this section we describe our visualization design

choice. Then, present the study we ran in VR, followed

by an overview on quantitative and qualitative results.

4.1 Visualization Design

We start by explaining the correlation between HR

data and ﬁtness activity monitoring. Then, present a

sample of existing HR data visualizations and explain

how they inspired our visualization design choice.

4.1.1

Heart Rate Zones to Monitor Sport Activity

During the course of a physical activity and depending

on the physical effort put into the exercise, the HR can

be in a speciﬁc zone. For example, during warm-up

or cool-down exercises, the HR is relatively low, and

Figure 3: HRZ representations from third-party applica-

tions: (a) horizontal progress bar from the Zones for Train-

ing app (Flask, 2018), (b) radial progress bar chart, and

itor app (FITIV, 2021), (d) radial gauge from the Google

Fit app (GoogleFit, 2020), and representations combining

HR and HRZ data in the same view from (e) the HRZMon+

app (MakkasCIQ, 2017): with a vertical and a progress bar

chart, and (f) the Cardiogram app (Cardiogram, 2017): show-

ing a bar chart encoding both data.

therefore, in the low intensity zone. A person’s range

of HRZs is based on the percentage of their maximum

HR, which is calculated based on the person’s age:

max HR = 220

−

age (American Heart Association,

2022) or = 205.8

−

0.685

∗

age (Friel, 2006) for more

accuracy. The number of HRZs can vary from three

(e. g., Google Fit and Cardiogram apps (Figure 3, d

and f)) up to seven zones (i. e., zone 5 is divided into

3 sub-zones w.r.t. the Lactate threshold, e. g. Heart

Rate Zones app (HC Studios, 2022)). For simpliﬁ-

cation purposes, we chose that our experiment’s HR

visualizations show only three HRZs. According to

the American Heart Association American Heart Asso-

ciation (2022), the three HRZ are deﬁned as low: HR

values

≤

50% of max HR, moderate: HR values are

50-85% of max HR, and high: HR values are

≥

85%

of max HR.

4.1.2 Existing Heart Rate and Heart Rate Zones

Visualizations

According to the answers that we gathered from the

survey participants on how they display their HR data

on their tracking devices during the course of an ex-

ercise, only eight reported visualizing “graphs/charts”

and “graph and text” representations. Among them,

three were using Garmin watches (1 Instinct and

2 Fenix 6X), two used Apple devices (Watch3 and

iPhone X), two had a Huawei Band 2 pro, and one an

AKWLOVY smartwatch. By investigating the visual-

IVAPP 2023 - 14th International Conference on Information Visualization Theory and Applications

106

Figure 4: Examples of HR visualizations that participants

reported to use during their exercise: (a) line chart visualiza-

tion on the Garmin Instinct smartwatch (Garmin, 2022), (b)

vertical bar chart visualization on the Garmin Fenix 6X (DC

RAINMAKER, 2019), and (c) vertical range chart on the

Apple Watch 3 (Apple, 2022).

izations provided by the in-house HR applications on

the aforementioned smartwatch models and other pop-

ular brands on the smartwatch market (statista, 2021)

(e. g., Samsung, Fitbit), we could identify three dis-

tinct types of HR visualizations, namely, line chart,

vertical bar chart, and vertical range chart (Figure 4).

These visualizations are often accompanied by textual

information on the minimum, maximum, average, or

resting HR values. Whereas information on the HRZs

is not by default available in all HR tracking applica-

tions. Therefore, to track the exercise time distribution

over HRZs in real time, smartwatch wearers use third-

party applications. We explored some of these ap-

plications and grouped the most commonly provided

visualizations into ﬁve types (i. e., linear and radial

progress bar, linear and radial gauge, and combined

visualizations (Figure 3)). Progress bar visualizations

(Figure 3, a and b) give details about the degree of

progress achieved per individual HRZ (i. e., in percent-

age, or seconds/minutes/hours). In contrast, Gauge

visualizations only indicate the current HR value and

highlight the current HRZ by fading out the unreached

or the rest of the HRZs (Figure 3, c and d). We noticed

that only two applications show visualizations of both

data simultaneously—allowing direct access to the in-

formation (i. e., the wearer does not have to navigate

between views or apps).

The HRZMon+ app (Figure 3, e) provides HR and

HRZ data using a bar chart and vertical progress bar,

respectively. Also, similar to the visualization given

by the Garmin Fenix 6X (Figure 4, middle), the Car-

diogram app by Apple watch (Figure 3, f), combines

the HR measurements over time—shown as a vertical

bar chart—to the HRZ information—represented by

the color of the bars.

Existing examples and recommendations from the

literature inspired the design choice of our HR visual-

izations. First, our goal was to test how well complex

visualizations (i. e., visualizations showing interrelated

data) can ﬁt the small screen of a smartwatch. Then, to

empirically assess them, by the means of an analytic

task: how fast these visualizations can convey precise

information to help a person meet a preset goal by

adjusting their performance in real-time.

4.1.3 Visualizations Design Choice

We created three visualizations showing the same in-

formation, (i. e., current and historical—since the be-

ginning of the exercise—HR, and HRZ data), yet with

distinct representations. The visualization choice was

mainly guided by the two design considerations from

the literature. First, all HR representations must be

charts. Then, convey information about interrelated

data (i. e., HR measures over time and time spent per

HRZ) to help prompt reﬂection and decision making

on the go. Therefore, the ﬁrst criterion that we want to

explore is whether combining the data into a single vi-

sualization (1) or showing them separately in the same

view (2) inﬂuences participants’ performance. Hence,

we opted for a bar chart visualization—rather than a

line or range chart—to depict HR data for two main

reasons. First, bar charts have the privilege of visually

encoding both the HR measures and zones simultane-

ously. For each HR measure the bar length depicts

the average HR value per minute and the color-coding

refers to the HRZ, to which the HR value belongs: yel-

low for low HRZ, orange for moderate HRZ, and red

for high HRZ (Figure 6). Second, with an overview

of the historical data, one can quantify and proportion-

ally estimate the time (in minutes) spent per HRZ, by

counting or visually grouping bars of the same color.

This makes the bar chart visualization an adequate

candidate for visualization category (1).

Accordingly, all three visualizations show as the

main chart a vertical bar chart of HR measures over

time. The current HR value is shown both as the most

right bar of the bar chart and as a text value displayed

at the top-center of the screen (Figure 6). Micro vi-

sualizations are characterized by an absence or a lim-

ited set of reference structures (i. e., labels, data axes).

Therefore, our HR visualizations show no x- or y-axis.

However, the x-axis shows the duration of the activity—

a maximum of 21 minutes—and the y-axis represents

the HR values in a range of 50–190 BPM.

For the visualization category (2), we display the

amount of time—in percentage—spent in each of the

three HRZ with a separate summary chart. We propose

two distinct representations for the summary chart

(i. e., horizontal and radial bar chart summaries, respec-

tively; Figure 6 (b) and Figure 6 (c)). The summary

chart is divided into three equal parts for low, moder-

ate, and high-intensity HRZ, ﬁlled with yellow, orange,

and red colors respectively, to indicate the percentage

of progress on each HRZ toward a pre-set goal. The

gray area depicts the remaining percentage to achieve

Heart Rate Visualizations on a Virtual Smartwatch to Monitor Physical Activity Intensity

107

Figure 5: A participant performing the study task: (a) the participant catches the virtual balls, (b) the participant looks at

the visualized HR data on the virtual smartwatch, (c) the participant validates his decision regarding the exercise pace (i. e.,

decrease: down-arrow, continue: circle, increase: up-arrow).

Figure 6: Examples of the HR visualizations from our study:

HR visualization—represented by a vertical bar chart show-

ing current and previous HR values—(a) without summary

chart, (b) HR visualization with horizontal bar chart sum-

mary, and (c) HR visualization with radial bar chart summary.

The horizontal and radial bar chart summary describes the

amount of time spent in each HRZ: low intensity: yellow,

moderate intensity: orange, high intensity: red.

during the exercise. When designing the summary

charts, we ensured that both the horizontal bar and

radial bar occupy the same amount of pixels on the

watch face. The horizontal summary chart mimics a

stacked bar chart, whereas the radial chart summary

better ﬁts the screen of the round (virtual) smartwatch.

For the sake of the study and to have an evenly

distributed dataset, we generated 27 unique stimuli

and used them for all three visualizations. We gener-

ated groups of nine datasets, which all had the same

current HRZ (i. e., 9 datasets for which the current

HRZ was in the low zone, 9 with a current moderate

HRZ, and 9 with a current high HRZ). For each group,

we sub-grouped the datasets to have a different HRZ

containing the minimum percentage of time spent in it

(i. e., an equal number of stimuli, in which the progress

of the current HRZ is the lowest or the highest). For

these three stimuli, we generated two datasets, which

represent a more difﬁcult decision and one dataset that

was an easier decision. The difﬁcult decision dataset

had a difference in number of bars of 1-2 between

the two HRZs with the least progress. Whereas the

easier decision dataset had a larger difference gap of

4-5 bars–which makes it easier to detect. For example,

Figure 6 (a), and Figure 6 (c) show a difﬁcult decision,

Figure 6 (b) shows an easy decision.

4.2 Study Task

Fitness Goal. We deﬁned a ﬁtness goal and informed

the participants about it before starting the game. The

goal was as follows: during an exercise session of 21

minutes in total, they have to try to be in each HRZ

equally—optimally 7 minutes per HRZ. To make a

decision, participants had to determine from the dis-

played visualization the HRZ during which they had

spent the least amount of time during the activity so

far. This can be concluded in two different ways: in

the summary chart ﬁnding the HRZ with the smallest

colored bar or by counting and comparing the bars of

each HRZ of the HR visualization. Then, they had to

identify the current HRZ, which can be determined by

the most right bar in the bar chart. Knowing their cur-

rent HRZ and the HRZ with the least global progress,

the participant should decide whether to decrease, con-

tinue with or increase the activity pace. For example,

the stimuli without summary chart (Figure 6 (a)) has

the high HRZ as the current HRZ and the least time

was spent in the moderate HRZ. Therefore, to balance

the distribution of time for each HRZ the decision in

this case would be to decrease the exercise intensity.

The decision for the stimuli with bar chart summary

(Figure 6 (b)) is to increase the intensity and for the

stimuli with radial bar chart summary (Figure 6 (c))

the decision is to continue with the intensity.

Primary and Secondary Task. During the study, par-

ticipants are constantly switching between two tasks.

The ﬁrst task involves action and movement. In this

IVAPP 2023 - 14th International Conference on Information Visualization Theory and Applications

108

primary task, the participant plays a tennis-like game

in VR by catching balls from both the left and right

side using virtual rackets (Figure 5 (a)). The secondary

task requires more reﬂection and concentration. The

secondary task consist of viewing the HR visualization

displayed on the virtual smartwatch (Figure 5 (b)), re-

ﬂecting on the decision to make to reach the preset goal

and ﬁnally validating their answer by selecting one of

the three buttons (i. e., down-arrow for decrease, circle

for continue, and up-arrow for increase; Figure 5 (c)).

Participants were given 10 seconds to make a deci-

sion. If no answer was selected within this period, the

answer was counted as incorrect.

4.3 Study Design and Apparatus

We conducted a within-subject study varying the type

of visualization shown to participants. Overall, we

had three conditions and one task resulting in a 3

1 study design. Each participant performed all three

conditions of the study. To prevent an ordering effect,

we used a 3

3 Latin Square to counterbalance the

conditions between participants.

We implemented a tennis-like game in VR using

the Unity3D game engine. We ran the VR game on a

Razer-Blade-15-Studio computer (i. e., CPU 8-Core In-

tel Core i7, GPU Quadro RTX™ 5000) and an Oculus

Rift VR headset. We cleared a space of approx. 1.5 m

3 m in the study room, so participants could move

freely when playing the game. To minimize the risk

of bumping into an object (i. e., walls, tables), we set

up the room boundary in VR (i. e., displaying a virtual

wired wall when getting too close to a real wall).

4.4 Study Procedure

We provided participants with a printed version of the

consent form. After signing the consent form, we

introduced them to the task via a short presentation.

Then, we showed participants an example of each of

the HR visualizations and asked them questions to

test their understanding of the task. Once explaining

the theoretical part was done, we asked participants

to wear the VR headset and to adjust it to their head.

Then, to ﬁnd the corresponding lens distance to their

inter-pupillary distance (IPD), they had to move the

IPD slider until the visuals looked clear.

Each condition was composed of two sessions: ﬁrst

the training session to get familiar with the VR envi-

ronment and visualizations, then the actual sessions.

Participants were exposed to 3 training stimuli and 27

visualization stimuli during each session. In total, each

participant performed 90 trials (i. e., 3

30), only 81

trials (i. e., 3

27) are considered for data analysis.

Between conditions, participants could take breaks.

Each session started when the participant pressed

the “start” button. A number of balls are shot toward

the player from both the left and right side randomly.

We asked participants to catch as many balls as pos-

sible using two rackets attached to each virtual hand

(Figure 5 (a)). To make the switch between the pri-

mary and the secondary tasks unpredictable for the

participants (i. e., checking data on the ﬁtness tracker

does not have to happen on a regular basis), the ball

stream stopped after a random number of balls was

shot (i. e., 2-7 balls). Then a question appeared in front

of the player asking “What would be your decision?”

Participants had to look at the virtual smartwatch (Fig-

ure 5 (b)), ﬁnd the answer from the HR visualization

displayed on the watch and validate their answer by

selecting one of the buttons shown below the question

(i. e., decrease: down-arrow, continue: circle, increase:

up-arrow; Figure 5 (c)). From the moment the question

appeared until participants pressed the answer button

a 10 seconds time loader kept counting down. If no

button was selected within the 10 seconds the answer

was considered as incorrect.

After participants preformed all 90 trials, we asked

them to ﬁll out a questionnaire to evaluate the three

HR visualizations. Participants who completed the

study received

10 or chocolate for their participation

if they were a university employee.

4.5 Participants

We recruited participants via the university’s internal

mailing list and student forums. In total, we recruited

24 participants: 20 students and 4 researchers of the

university, including 11 females and 13 males. All par-

ticipants were 18 years old or older. They conﬁrmed

that they have good eyesight (i. e., their vision is good

without requiring an optical aid or is sufﬁciently cor-

rected by means of an optical aid) and that they do

not have a physical injury that constrains them from

moving their arms freely.

4.6 Results

We recorded data logs from participants with measures

on their performance for each condition, as well as

subjective feedback at the end of the study. In this

section, we report quantitative and qualitative results.

4.6.1 Quantitative Results

We analyze the mean response time and the mean

accuracy per condition (i. e., 648 trials (27 trials

participants) per condition) using conﬁdence interval

Heart Rate Visualizations on a Virtual Smartwatch to Monitor Physical Activity Intensity

109

Figure 7: Mean completion time (left) and difference between mean completion times (right) of the three HR visualizations

using 95% CIs and Bonferroni corrections.

Figure 8: Mean accuracy (left) and difference between mean accuracy (right) of the three HR visualizations using 95% CIs and

Bonferroni corrections.

(CI) estimation (Besan

c¸

on and Dragicevic, 2017, 2019;

Cockburn et al., 2020; Cumming, 2013; Dragicevic,

2016). We report the sample means of the completion

time and accuracy together with the 95% CI using

BCa bootstrapping (10,000 bootstrap iterations). In

addition, we report the differences between means to

compare the three conditions. We adjusted the CIs

of mean differences for multiple comparisons with

Bonferroni corrections (Higgins, 2004).

On average, participants were faster with the radial

bar chart summary (3929 ms [3637 ms, 4211 ms]) than

with the other conditions (horizontal bar chart sum-

mary: 4038 ms [3786 ms, 4389 ms], without summary

chart: 4462 ms [4173 ms, 4763 ms]) (Figure 7 left).

The results show evidence of a difference in terms of

completion time (Figure 7 right), for the radial bar

chart summary and the without summary chart (CI

does not touch zero: 532 ms [263 ms, 874 ms]) and a

trend for a difference between the horizontal bar chart

summary in comparison to the without summary chart

(CI is close to zero: 423 ms [91 ms, 767 ms]). There

is no difference in mean completion time between the

radial and horizontal bar chart summary (CI crosses

over zero: 109 ms [-155 ms, 380 ms]).

Mean accuracy results show that participants’

made more accurate decisions with the horizontal bar

chart summary (92.7% [87%, 95.8%), than the radial

bar chart summary (92.3% [84.2%, 96.2%) and the

without summary chart (83.0% [75.6%, 88.4%) (Fig-

ure 8 left). There is evidence of a difference for both

the radial and horizontal bar chart summary to the

without summary chart (-9.2% [-15.7%, -5.2%]) and (-

9.7% [-16.8%, -4.6%]); Figure 8 right). There is no ev-

idence of a trend between the radial and the horizontal

chart summary (0.4% [-3.3%, 6.7%]); Figure 8 right).

Figure 9: Participants’ self-assessment about their conﬁ-

dence (top) and how helpful the three visualizations were

(bottom), in answering the task. (a) without summary, (b)

with bar chart summary, (c) with radial chart summary.

4.6.2 Qualitative Results

After performing the study with all three visualizations,

we asked participants to answer three questions to

evaluate their satisfaction with each visualization.

1) How Do You Evaluate Your Conﬁdence in An-

swering the Task with the Three Visualization?

Results in Figure 9 (top) depict participants’ self-

assessment of the degree of conﬁdence in their an-

swers with each visualization. For the visualization

not containing a summary, three participants stated

that they were not conﬁdent at all about their answers

IVAPP 2023 - 14th International Conference on Information Visualization Theory and Applications

110

—which no participant reported for the two other visu-

alizations showing a summary chart. In addition, only

four participants found that they were conﬁdent in

their answers using the visualization without summary

chart compared to 21 participants (i. e., 10 conﬁdent

and 11 very conﬁdent) for the visualization with bar

chart summary and 22 participants (i. e., 5 conﬁdent

and 17 very conﬁdent) for the visualization with radial

chart summary.

Some participants explained this gap in the results

by the fact that when the visualization does not show

a summary they had to calculate and visually compare

the number of bars in each HRZ—which was tedious

and time-consuming. For example, one participant

stated: “The summary makes me feel more conﬁdent

about my choice given that I didn’t take a lot of time

to think about it.”

2) How Helpful Were the Visualizations in Making

a Decision? Only ﬁve participants found the visual-

ization without summary helpful for the task (Figure 9,

bottom). In contrast, 21 participants agreed that the

bar chart summary (i. e., 8 helpful and 13 very helpful)

and radial chart summary (i. e., 3 helpful and 18 very

helpful) were beneﬁcial to solve the task.

3) Which Visualization Did You Prefer the Most?

Finally, we asked participants to choose the visual-

ization they preferred the most based on aesthetics,

efﬁciency and conﬁdence. The majority of the par-

ticipants (83%) preferred the visualization with the

radial chart summary, 17% opted for the visualization

with the bar chart summary, and none voted for the

visualization without summary.

5 DISCUSSION

In general, participants could accomplish the task suc-

cessfully with the three visualizations by providing

quick responses (i. e.,

5 s) and with most of the an-

swers being correct (83%-92.7% on average). How-

ever, when interrelated HR data was depicted using

two visualizations (i. e., the vertical bar chart and the

summary chart), participants’ performance increased.

A particular preference was accorded to the visualiza-

tion with the radial chart summary (w.r.t. aesthetics,

efﬁciency, and conﬁdence in the provided answers).

In this last section, we discuss our ﬁndings and

highlight some advantages and limitations that we

identiﬁed from our results or participants’ feedback.

5.1 Exploring Visualizations on Small

Screens While Exercising (RQ1)

Smartwatches are usually underestimated for the

amount and quality of information they can provide

because of their restricted display area. Previous

work (Islam et al., 2020) and our own survey show

that most data is shown as text, icon or as a combi-

nation of text and icon on the smartwatch. However,

representing raw textual information does not allow

for deeper reﬂection when exercising (Amini et al.,

2017; Neshati et al., 2019a). Therefore, visualizing the

data can overcome this challenge, but requires careful

design of these visualizations to ﬁt the small screen.

By following design recommendations from previous

works (Amini et al., 2017; Neshati et al., 2019a; Choe

et al., 2017), we opted for the use of simple and fa-

miliar visualizations (e. g., bar and radial charts) as a

replacement for raw textual data, to support reﬂection

in a goal-oriented task.

Our ﬁnal results showed that the small screen of a

(virtual) smartwatch can convey informative feedback

with minimum interruption from the principal activ-

ity in a short amount of time (

5s), which is in line

with previous studies (Blascheck et al., 2019; Neshati

et al., 2021a, 2019b). This is even true, when people

perform a primary task (playing a tennis-like game).

The recourse to simple visualizations on smartwatch

screens has a positive impact on the reﬂection time

and outcome in general, especially for tasks performed

while in motion or that involve task switching.

5.2 Multiple Visualizations to Support

Complex Data Interpretation (RQ2

& RQ3)

According to our survey and other previous works (Is-

lam et al., 2020; Choe et al., 2017), people tend to

track multiple types of data at the same time during

their activities. However, if no clear relation exists

between the data, combining multiple data types—

especially if using different visualization techniques—

simultaneously can make data interpretation a complex

process (Neshati et al., 2019a; Choe et al., 2017). In

this study, we focused on visualizing interrelated data

types to serve answering a common question, namely,

HR measures over time and the total amount of time

per HRZ. All three of our visualizations showed the

same information but differently. In the visualization

without a summary, it was possible to estimate the total

amount of time per HRZ by summing bars with the

same color. However, this was time consuming and led

to more errors of interpretation. Furthermore, many

Heart Rate Visualizations on a Virtual Smartwatch to Monitor Physical Activity Intensity

111

participants reported that the task was mentally tiring

and they felt uncertain about their answer. For exam-

ple, one participant mentioned: “In the visualization

without summary I always was not sure if I have not

missed anything” another stated that “it was difﬁcult

to solve the task without the summary.” Accordingly,

we assume that the task would be harder to handle for

a longer exercise period (i. e., >21 minutes).

Our results show that providing the total amount

of time per HRZ using a separate visualization (i. e.,

by adding a horizontal or radial summary chart) made

the data processing and the interpretation of the data

easier and more efﬁcient (e. g., “having a summary is

a nice extra feature that is deﬁnitely helpful”). This

opens up new research questions on the impact of data

aggregation on different tasks that require reﬂection

and with different data granularity (i. e., more than

two data). Also, the maximum number of visualiza-

tions a smartwatch can display simultaneously and

still convey meaningful and clear information could

be inspected.

5.3 Monitoring a Physical Activity in

Virtual Reality (RQ4)

This study had the ultimate goal of simulating multi-

ple tasks (i. e., primary and secondary task) with the

secondary task requiring a participant to read data on

a virtual smartwatch. For this, we designed visualiza-

tions and displayed them on the virtual smartwatch

simulating the size of a standard real watch. In the

following, we discuss how a virtual smartwatch can

simulate a real watch to study micro visualizations.

Rendering Quality. In general participants found the

visualizations’ rendering clear enough (i. e., partici-

pants were able to extract the necessary information

without problems). Some participants mentioned that

“the rendering quality seemed to be equally good” or

“was good, easy to read” when we asked them to evalu-

ate the rendering clarity. However, some reported that

the text value indicating the current HR measure was

blurry, which made it hard to read. For the proposed

task, this information was not crucial to answering

the question. We suggest that future works involving

micro visualizations in VR do not or only sparsely in-

clude text-based representation on the watch to avoid

alteration of the texts’ quality.

Visual Jitter. Although, when retrieving information

from the visualization on the smartwatch, participants

were not moving, they performed a quick arm-twist

to look at the virtual smartwatch. One participant re-

ported that the speed at which his arm moved affected

the clarity of the visualizations for a short time (i. e.,

the faster, the more blurred the representation). We

do not consider this effect as a problem. First because

it is brief; second, because it simulates the effect of

jitter that can also occur when visualizing data on a

real smartwatch.

Restricted Displacement. On the one hand, running

the study in a controlled VR environment helped us

to quantify participants reactions (i. e., response time

and accuracy). On the other hand, it was limiting par-

ticipants in terms of displacement space. We could

not risk participants’ safety by testing our study setup

with exercise activities that involve high motion in a

real environment (e. g., walking or running) or contin-

uous displacement (i. e., in the room or on a treadmill).

Therefore, the VR environment was a good trade-off

between realism and safety of participants.

6 CONCLUSION

In this paper, we report the results and ﬁndings from

our survey and VR study. We created three HR visual-

izations guided by our survey’s ﬁndings and by design

considerations from the literature that support visual

data processing and reﬂection on the go. Through a

study, we demonstrated the efﬁciency of the three vi-

sualizations displayed on small wearable devices in

guiding a person to make decisions about their exercise

pace. We also found that data aggregation impacted

the performance of the participants. The HR visual-

izations showing a summary of the total time spent

in each HRZ represented with a separate radial and

horizontal bar chart summary were more helpful to per-

form the task. Participants preferred the visualization

with radial bar chart summary the most.

Overall, the obtained results are promising and en-

courage us to further explore in-situ visualizations

in situations that require quick reﬂection in gen-

eral, and in the context of ﬁtness data tracking more

speciﬁcally—whether by means of a real or a virtual

smartwatch. A combination of both modalities can be

considered using Augmented Reality.

A possible extension of this work would be to test

the same or similar visualizations with different data

granularities. In addition, testing the HR visualizations

with real HR data would further show how usable the

visualizations are in a real-world context and in situ-

ations where the attention is split between a primary

and a secondary tasks.

ACKNOWLEDGEMENTS

We thank all the participants who took part in our

study. This project is funded under DFG grant

IVAPP 2023 - 14th International Conference on Information Visualization Theory and Applications

112

DFG ER 272-14 and by the Deutsche Forschungs-

gemeinschaft (DFG, German Research Foundation)

under Germany’s Excellence Strategy – EXC 2075

– 390740016. Tanja Blascheck is funded by the Eu-

ropean Social Fund and the Ministry of Science, Re-

search and Arts Baden-W

urttemberg.

REFERENCES

Albaghli, R. and Anderson, K. (2016). A vision for heart

rate health through wearables. In Proceedings of the

ACM International Joint Conference on Pervasive and

Ubiquitous Computing: Adjunct, pages 1101–1105.

ACM.

American Heart Association (2022). Target heart rates

chart. https://www.heart.org/en/healthy-living/ﬁtness/

ﬁtness-basics/target-heart-rates. Last visited: October,

2022.

Amini, F., Hasan, K., Bunt, A., and Irani, P. (2017). Data

representations for in-situ exploration of health and

ﬁtness data. In Proceedings of the EAI International

Conference on Pervasive Computing Technologies for

Healthcare, pages 163–172. ACM.

Apple (2022). Check your heart rate on apple

watch. https://support.apple.com/guide/watch/

heart-rate-apda88aefe4c/watchos. Last visited:

October, 2022.

Besan

c¸

on, L. and Dragicevic, P. (2017). The signiﬁcant

difference between p-values and conﬁdence intervals.

In Proceedings of the Conference on l’Interaction

Homme-Machine, pages 53–62.

Besan

c¸

on, L. and Dragicevic, P. (2019). The continued

prevalence of dichotomous inferences at CHI. In Ex-

tended Abstracts of the SIGCHI Conference on Human

Factors in Computing Systems, pages 1–11. ACM.

Blascheck, T., Bentley, F., Choe, E. K., Horak, T., and Isen-

berg, P. (2021). Characterizing glanceable visualiza-

tions: from perception to behavior change. In Lee, B.,

Dachselt, R., Isenberg, P., and Choe, E. K., editors,

Mobile Data Visualization, chapter 5, pages 151–176.

Chapman and Hall/CRC.

Blascheck, T., Besan

c¸

on, L., Bezerianos, A., Lee, B., and

Isenberg, P. (2019). Glanceable visualization: stud-

ies of data comparison performance on smartwatches.

IEEE Transactions on Visualization and Computer

Graphics, 25(1):630–640.

Blascheck, T. and Isenberg, P. (2021). A replication study on

glanceable visualizations: Comparing different stimu-

lus sizes on a laptop computer. In VISIGRAPP.

Cardiogram, H. R. M. (2017). https://apps.apple.com/

app/cardiogram-heart-rate-monitor/id1000017994?

platform=appleWatch. Last visited: October, 2022.

Chen, Y. (2017). Visualizing large time-series data on very

small screens. In Short Papers of the European Confer-

ence on Visualization, pages 37–41. The Eurographics

Association.

Choe, E. K., Lee, B., Zhu, H., Riche, N. H., and Baur, D.

(2017). Understanding self-reﬂection: how people

reﬂect on personal data through visual data exploration.

In Proceedings of the EAI International Conference

on Pervasive Computing Technologies for Healthcare,

pages 173–182. ACM.

Cockburn, A., Dragicevic, P., Besan

c¸

on, L., and Gutwin,

C. (2020). Threats of a replication crisis in empiri-

cal computer science. Communications of the ACM,

63(8):70–79.

Cumming, G. (2013). Understanding the new statistics:

effect sizes, conﬁdence intervals, and meta-analysis.

Routledge.

DC RAINMAKER (2019). Garmin fenix 6 series in-

depth review. https://www.dcrainmaker.com/2019/

08/garmin-fenix6-pro-solar-series-review.html/

comment-page-2. Last visited: October, 2022.

Dragicevic, P. (2016). Fair statistical communication in HCI.

In Robertson, J. and Kaptein, M., editors, Modern

Statistical Methods for HCI, chapter 13, pages 291–

330. Springer.

FITIV, M. F. I. (2021). Pulse heart rate monitor. https:

//ﬁtiv.com/. Last visited: October, 2022.

Flask, L. (2018). Zones for training. https:

//apps.apple.com/us/app/zones-for-training/

id1139688415?platform=appleWatch. Last vis-

ited: October, 2022.

Friel, J. (2006). Total Heart Rate Training. Ulysses Press.

Garmin (2022). Garmin instinct heart

rate widget. https://www8.garmin.

com/manuals/webhelp/instinct/EN-US/

GUID-EB942FCB-6DA6-4082-9ED3-0DE51CB79E74.

html. Last visited: October, 2022.

GoogleFit (2020). Updates to google ﬁt on wear os.

https://support.google.com/wearos/thread/84883193/

updates-to-google-ﬁt-on-wear-os?hl=en. Last visited:

October, 2022.

HC Studios (2022). Heart rate zones app.

https://play.google.com/store/apps/details?id=

com.hcstudios.heartratezonecalculator. Last visited:

October, 2022.

Higgins, J. (2004). Introduction to Modern Nonparametric

Statistics. Thomson Learning.

Islam, A., Aravind, R., Blascheck, T., Bezerianos, A., and

Isenberg, P. (2022). Preferences and effectiveness of

sleep data visualizations for smartwatches and ﬁtness

bands. In Proceedings of the SIGCHI Conference on

Human Factors in Computing Systems, pages 1–17.

ACM.

Islam, A., Bezerianos, A., Lee, B., Blascheck, T., and Isen-

berg, P. (2020). Visualizing information on watch faces:

a survey with smartwatch users. In Short Papers of the

IEEE Visualization Conference, pages 156–160. IEEE

Computer Society Press.

Ketcheson, M., Ye, Z., and Graham, N. (2015). Designing for

exertion: how heart-rate power-ups increase physical

activity in exergames. In Proceedings of the Annual

Symposium on Computer-Human Interaction in Play,

pages 79–89. ACM.

MakkasCIQ (2017). Hrzmon+, heart rate zone

monitor+. https://apps.garmin.com/en-US/apps/

Heart Rate Visualizations on a Virtual Smartwatch to Monitor Physical Activity Intensity

113

79d55484-c0d2-4047-b7ee-72d051d86e14. Last vis-

ited: October, 2022.

Muangsrinoon, S. and Boonbrahm, P. (2017). Burn in zone:

real time heart rate monitoring for physical activity. In

International Joint Conference on Computer Science

and Software Engineering, pages 1–6. IEEE Computer

Society Press.

Navarro, P., Johns, M. L., Lu, T.-H., Martin, H., Poduval,

V., Robinson, M., Roxby, A., and Christel, M. (2013).

Webz of war: a cooperative exergame driven by the

heart. In IEEE International Games Innovation Confer-

ence, pages 187–190. IEEE Computer Society Press.

Neshati, A., Alallah, F., Rey, B., Sakamoto, Y., Serrano,

M., and Irani, P. (2021a). SF-LG: space-ﬁlling line

graphs for visualizing interrelated time-series data on

smartwatches. In ACM MobileHCI, pages 1–13. ACM.

Neshati, A., Rey, B., Mohommed Faleel, A. S., Bardot, S.,

Latulipe, C., and Irani, P. (2021b). BezelGlide: interact-

ing with graphs on smartwatches with minimal screen

occlusion. In Proceedings of the SIGCHI Conference

on Human Factors in Computing Systems, pages 1–13.

ACM.

Neshati, A., Sakamoto, Y., and Irani, P. (2019a). Challenges

in displaying health data on small smartwatch screens.

In Lau, F., Bartle-Clar, J., Bliss, G., Borycki, E., Court-

ney, K., Kuo, A. M.-H., Kushniruk, A., Monkman, H.,

and Roudsari, A., editors, Improving Usability, Safety

and Patient Outcomes with Health Information Tech-

nology, pages 325–332. IOS Press.

Neshati, A., Sakamoto, Y., Leboe-McGowan, L., Leboe-

McGowan, J., Serrano, M., and Irani, P. (2019b). G-

Sparks: glanceable sparklines on smartwatches. In Pro-

ceedings of the Graphics Interface Conference, pages

1–9. Canadian Human-Computer Communications So-

ciety.

OSF (2022). Study supplemental mate-

rial. https://osf.io/rbg64/?view only=

1dce5b18784b432ba1557d7e5a04e81d. Last

visited: December, 2022.

Pew Research Center (2022). About one-in-ﬁve

americans use a smart watch or ﬁtness tracker.

https://www.pewresearch.org/fact-tank/2020/01/09/

about-one-in-ﬁve-americans-use-a-smart-watch-or-ﬁtness-tracker/.

Last visited: October, 2022.

Queck, D., Albert, I., Burkard, N., Zimmer, P., Volkmar, G.,

anekas, B., Malaka, R., and Herrlich, M. (2022). Spi-

derclip: towards an open source system for wearable

device simulation in virtual reality. In Extended Ab-

stracts of the SIGCHI Conference on Human Factors

in Computing Systems, pages 1–7. ACM.

Schiewe, A., Krekhov, A., Kerber, F., Daiber, F., and Kr

uger,

J. (2020). A study on real-time visualizations during

sports activities on smartwatches. In International Con-

ference on Mobile and Ubiquitous Multimedia, pages

18–31. ACM.

She, H., Young, J., Wittenberg, O., Poole, P., Mercer, A.,

Shaw, A., and W

unsche, B. (2020). HIITCopter: anal-

ysis of the suitability of VR exergaming for high-

intensity interval training. In Australian Conference on

Human-Computer Interaction, pages 293–302. ACM.

Siani, A. and Marley, S. (2021). Impact of the recreational

use of virtual reality on physical and mental wellbeing

during the covid-19 lockdown. Health and Technology,

11(2):425–435.

SosciSurvey (2022). Online survey tool. https://www.

soscisurvey.de. Last visited: October, 2022.

statista (2021). Market share of smartwatch between 2014-

2021, by vendor. https://www.statista.com/statistics/

524830/global-smartwatch-vendors-market-share/.

Last visited: October, 2022.

Suciu, A. and Larsen, J. (2018). Active self-tracking and

visualization of subjective experience using VAS and

time spirals on a smartwatch. In Workshop on Data

Visualization on Mobile Devices held at ACM CHI.

Tufte, E. (2001). The Visual Display of Quantitative Infor-

mation. Graphics Press.

Yao, L., Bezerianos, A., and Isenberg, P. (2020). Situated

visualization in motion. In Posters of the IEEE Visual-

ization Conference.

Yoo, S., Parker, C., and Kay, J. (2018). Adapting data

from physical activity sensors for visualising exertion

in virtual reality games. In Proceedings of the ACM

International Joint Conference and International Sym-

posium on Pervasive and Ubiquitous Computing and

Wearable Computers, pages 307–310. ACM.

YUR.Watch (2020). Yur.watch. https://www.yur.energy/

#virtual-watch. Last visited: October, 2022.

IVAPP 2023 - 14th International Conference on Information Visualization Theory and Applications

114