ArticlePDF Available

Advances for Indoor Fitness Tracking, Coaching, and Motivation: A Review of Existing Technological Advances


Abstract and Figures

There is growing consumer demand for digital technologies that help users track, motivate, and receive coaching for both aerobic and anaerobic activities. In this article, we provide a review of existing technological advances in tracking, coaching, and motivating users during indoor training in contexts such as gymnasiums. This study lists the advantages and limitations of various apparatuses and applications used for this purpose. Our review and discussion are intended to help entrepreneurs and engineers improve their products to better meet users? needs and aid researchers in identifying potential new areas.
Content may be subject to copyright.
Digital Object Identifier 10.1109/MSMC.2020.3017936
Date of cur rent version: 14 Januar y 2021
There is growing consumer demand for digital
technologies that help users track, motivate,
and receive coaching for both aerobic and
anaerobic activ ities. In this article, we pro-
vide a review of existing technologica l
advances in tracking, coaching, and motivating users
during indoor training in contexts such as gymnasiums.
This study lists the advantages and limitations of various
Advances for
Indoor Fitness
Tracking, Coaching,
and Motivation
by Tianyi Wang, Yanglei Gan,
Scott D. Arena, LubomirT.Chitkushev,
Guanglan Zhang, and
Reza Rawassizadeh
A Review of Existing Technological Advances
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
apparatuses and applications used for this purpose. Our
review and discussion are intended to help entrepreneurs
and engineers improve their products to better meet
users’ needs and aid researchers in identifying potential
new areas.
Indoor Physical Activities
Due to the rapid growth of urbanization, many populations
participate in sports and training in indoor environments,
chiefly gyms. Physical activ ity fosters well-being and
reduces the risk of various chronic
diseases. As cases of diabetes [1]
and cardiovascular disease [2] are
on the rise worldwide, the demand
for gyms has continually increased.
The U.S. fitness industry is estimat-
ed at a US$30 billion value and has
been growing at an annual rate of
3–4% over the past 10 years [3].
This demand creates a large mar-
ket for digital apparatuses and
applications that assist users in
fitness tracking, coaching, and
motivation [4].
Indoor activities are usually
classified into two types: aerobic and anaerobic exercise.
Aerobic exercise, often known as cardio, is suitable for
users who want to lose weight and improve or maintain
their cardiopulmonary condition. Modern gyms and
homes use equipment such as treadmills, stair mills, row-
ing machines, and spin bikes, which support a variety of
aerobic exercise. Anaerobic exercise is primarily accom-
plished with weight training activities, which contribute
to training specific muscles. The U.S. Department of
Health and Human Services recommends incorporating
strength training exercise for all ma jor muscle groups into
a fitness routine at least twice per week [3], indicating
that it is as impor tant as aerobic exercise.
In par ticular, there are two types of equipment used
in the gym that apply to weight training: free weights
and fixed resistance. Despite the wide variety of technol-
ogies related to indoor physical activities, most existing
technologies are focused on fitness trackers and aerobic
exercise [5], while anaerobic exercise, such as weight
training, is underrepresented [6]. Therefore, we believe it
is crucial to outline the latest advancements related to
indoor fitness and list the potential areas for further
improvement to contribute to the research and engineer-
ing communities’ work in this area. To have a better
understanding of the challenges and opportunities in
this area, the first par t of this article introduces existing
systems. First, we describe data collection tools a nd
methods, including applications and sensors. Next, we
present different approaches to motivate users to per-
form and stay committed to physical exercise. This
includes a brief review of the coaching tools that employ
artificial agents or real humans to coach users, social-
ization approaches that support users in communicating
and competing with each other, and gamification and
exergaming approaches intended to promote healthy
lifestyles and fitness activ ities. Afterward, we explore
the interaction interfaces used in these systems as a sub-
section of motivational advances.
The latter part of this article lists the challenges and
opportunities that we identified, which can help the
research and the engineering community move forward in
this promising direction, with an
eye toward further advancements.
Existing Systems
Data Collection Tools
Most technological advances have
focused on data collection. Here,
we provide a brief overview of the
types and characteristics of sen-
sors used for data collection.
Sensor Classification
The core of data collection devic-
es is their sensors, and existing
activ ity tracking systems can be broadly classified into
two categories: proximity [7] and mobile/wearable [8]
sensors. However, there is no clear border between the
categories, and it is possible for a single sensor to
belong to both [7]. At the time of writing this article,
wearable sensors are being commercialized and widely
available in the consumer market. On the other hand,
proximity sensors are in their infancy and not yet wide-
ly in use; we believe that they will soon be commercial-
ized by fitness service providers. However, unlike
wearables, which are personalized, they should be
shared by a specific group of users. Because wearable
devices are heterogeneous, for example, the recent Wear
operating system improves battery use compared to its
old version [9], and because proximity sensors are not
yet widely in use, werefrain from prov iding a quantita-
tive comparison in this research.
Proximity Sensors
A proximity (environmental) sensor is fixed and does not
move with the user [see Figure 1(a)]. In contrast, a mobile
sensor is wearable and will move with its user [see Fig-
ure 1(b)]. An example of a proximity sensor is a pulse
sensor (used to measure heart rate), which is mounted
on the handle of a device —for example, an elliptical—to
read a user’s pulse rate. This information will be dis-
played on the device’s ambient screen, which may also
show some basic coaching advice, such as a recommend-
ed maximum heart rate based on the age the user inputs,
or a suggestion of when to decrease the intensity of the
exercise. Heart rate sensors and ambient displays have
In particular, there
are two types of
equipment used in
the gym that apply
to weight training:
free weights and
fixed resistance.
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
Proximity Sensors Wearable Sensors
Connection (Count Reps)
Weight (Count Weights)
Gym Mat
Smart Textile/Short
Smart Textile/Shirt
Ear Forehead
Chest Strap/Necklace
Heart Rate
Pressure Sensor
Depth Camera Electromyography
Figure 1. The different locations of (a) proximity and (b) wearable sensors. reps: repetitions; RFID: radio-frequency identification.
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
been widely adapted in existing aerobic exercise
machines. The use of radio-frequency identification
(RFID) tags, pressure sensors, or accelerometers on
dumbbells has been commercialized as well (https://
Many researchers have developed systems that inte-
grate proximity sensors. For example, Chaudhri et al.
[10] and Ding et al. [11] introduced a system that employs
RFID: they attached RFID tags to equipment such as
dumbbells and fixed resistance machines to allow users
to track anaerobic exercise by connecting a tag reader
with the tag located on the device. Sundholm et al. [12]
presented a mat with a resistive pressure sensor matrix
built inside it that can distinguish gym exercises, such as
push-ups, abdominal crunches, squats, and so on.
To provide visual feedback on workout quality, Nagar-
koti et al. [13] proposed a system that can use a 2D ca mera
to detect users’ errors in workout form. Another body of
work employs depth cameras for activity tracking.
Although Microsoft stopped the production of Kinect sen-
sors in 2017, several in-depth camera approaches to indoor
exercise tracking inspired by the Kinect sensor have been
developed. For example, Khurana et al. [14] proposed a
Kinect sensor-based system that can detect, recognize, and
track exercise without user intervention. Lin et al. [15]
offered a method to monitor energy expenditure during
physical activity using Kinect sensors. Specifically, they
employed Kinect to collect skeletal data and then mapped
the data to a regression model. Another approach using
Kinect sensors, this one proposed by Reily et al. [16], can
evaluate users’ performance on the pommel horse bench—
measuring the consistency of the user’s timing and body
angle. Compared to a traditional gym-based exercise, a 3D
camera approach can enhance postural control and opti-
mize user experience [17].
A salient advantage of proximity sensors is their unob-
trusiveness. Users do not need to wear anything or bring
their own tracking devices. However, to archive their
information, they should transfer their data to personal
storage (e.g., a smartphone), which is not easy in proximity
to sensor devices. This is due to the fact that there is a bur-
den for users to connect the proximity device into their
personal storage; they must perform actions such as log-
ging into their account or establishing a Bluetooth connec-
tion with the device.
Wearable Sensors
About a decade ago, the success of activity tracking appli-
cations on smartphones [18]–[20] led to the introduction of
single-purpose wearable devices for activity tracking—fit-
ness trackers. The wide availability of fitness trackers in
the consumer market and the variety of wearable and
mobile sensors make them popular tools for the automatic
data collection of aerobic exercise. Wearable sensors are
typically embedded with accelerometers, gyroscopes, and
electrical and optical heart rate sensors and can actively
track a user’s heart rate along with instances of aerobic
exercise, such as cycling, running, and swimming [see Fig-
ure 1(a)]. Wearable sensors can communicate with ubiqui-
tous personal dev ices such as smartphones and construct a
wireless-body sensor network (BSN). Studies have shown
that many human-centric applications such as m-Health
and fitness/wellness systems benefit from BSNs [21]–[23].
Wrist-worn wearables are the most common type in
this category. For example, smartwatches are commer-
cialized and used widely. However, a recent study found
that ankle-worn wearables offer more reliable and sensi-
tive results than do wrist-worn wearables [24]. Remark-
able initial work from Nam and Park [25] proposed a
waist-worn system embedded with an accelerometer and
a barometric pressure sensor that can detect activ ities
such as climbing up and down and so forth. Iskandar et al.
[3] combined an electrocardiogram with a necklace to
determine a user’s heart rate. Cruz et al. [26] suggested
adopting an earpiece that uses infrared thermometry to
detect heart rate. Akpa et al. [5] offered a glove that can
track fitness activities when the user touches exercise
equipment; 16 force-sensitive resistor sensors on the glove
allow it to track anaerobic exercises such as flexibility
training (e.g., side lunge stretches), dynamic strength
training (e.g., squats), static strength tra ining (e.g.,
planks), and circuit training (e.g., push-ups, bench dips,
and lunges). Another force-sensitive approach proposed
by Zhou et al. [27] introduces a real-time assisted training
feedback system. Their system uses a smart textile on a
shirt as a fabr ic sensor [28] to track muscle activities by
measuring the pressure on the fabric. Using data acquired
from the shirt sensor, the system can classify anaerobic
exercise such as push-ups, bicep curls, handstands, and
so on. Textile wearables track anaerobic exercises by ana-
lyzing muscle responses via electromyography (EMG)
sensors during weight training activities. Another exam-
ple utilizing EMG was offered by Taha et al. [29]. They sug-
gested attaching an EMG sensor to the user’s bicep to
monitor exercise intensity and muscle fatigue.
There are few reports about the feasibility of wearing
gloves or adapting smart shir ts while exercising. In addi-
tion, in contrast to smartwatches and other wrist-worn
wearables, we have not obser ved a proliferation of smart
clothes in the wearable market. This might be due to their
high cost as well a s hygiene concer ns about the frequent
use of smart clothes. There is a need, however, for further
scientific investigation in this area to quantify the chal-
lenges that hinder acceptance of other wearables.
One advantage of using wearables for activity
tracking is their personalized information collection
approach. In other words, no effort from users is needed
to collect or retain their own data, in contrast to proxim-
ity sensors, where data collection devices are shared
among users. Table 1 summarizes some common data col-
lection approaches, based on the type of sensors, i.e.,
wearable and proximity settings.
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
Motivational Approaches
There are motivational applications intended to increase
user participation in physical activities, reduce their sed-
entar y behavior, or improve the quality of their physical
activ ities. Based on our analysis of the literature, we cat-
egorize motivational approaches into three groups:
coaching, socialization, and gamification/exergaming. To
our knowledge, there is little research done on motiva-
tional approaches. This limitation does not allow us to
generalize our findings and make a conclusive remark.
We believe that in the near future, more motivational
applications will enter the market because of artificial
intelligence (AI) advances.
In addition to activity data collection in fitness tracking
applications, users can interact with applications for anal-
ysis of the collected data. Early work in this area by Hiil-
loskorpi et al. [36] introduced a coaching approach widely
adopted by aerobic exercise devices: calculation of the
user’s maximum heart rate. The field of coaching applica-
tions is slowly advancing. Rawassizadeh et al. [37] suggest-
ed a coaching approach in which textual recommendations
are provided once a day because the small screen of most
wearable devices cannot accommodate graphs and num-
bers. Their approach compares users’ daily activities and
heart rate to their personal averages using a machine
learning algorithm and notifies them if their activity level
has decreased.
Another coaching approach proposed by Pessemier and
Martens [38] compares the data collected from heart rate
sensors mounted on exercise equipment or within wear-
able devices. Their use guides users in adjusting pace dur-
ing an aerobic exercise to avoid overexercise injuries or
more effectively reach their fitness goals.
Adaptation AI could improve fitness coaching applica-
tions considerably. Li et al. [39] proposed an AI fitness sys-
tem that can recognize fitness actions by applying
evaluation methods on image sequences. Furthermore, the
AI of Things (AIoT), which integrates AI, deep learning,
and the Internet of Things (IoT), is also advancing rapidly.
For example, Chu et al. [40] proposed an AIoT system that
consists of both biological IoT sensors on the body and
motion sensors on weight training equipment. Their sys-
tem can detect incorrect movements and provide coaching
advice to the trainee. Coaching approaches are still in
their infancy, and there is plenty of room for improvement,
as we explain later in detail.
Some other approaches focus on motivating users by shar-
ing their activity data through social network applications,
that is, through socialization. Studies have pointed out that
sharing physical activities can increase user motivation
[41], [42] by enabling users to compare their performance
with each other.
In this context, Häkkilä et al. [43] introduced a virtual
city jogging concept that simulates the subject’s move-
ment on a treadmill in a 3D virtual city. This application
combines a map-based presentation of the user’s jogging
distance and a 3D representation of the city. Users can
see previous joggers’ results and compete with them.
Cassola et al. [44] built an online gym system that satis-
fies users’ socialization and fitness needs from their
homes. The practice of employing socialization for moti-
vation is mature enough that it has penetrated the
mobile health market [e.g., Runkeeper (https://play
., Nike Run Club (
.com/store/apps/details?, and
Type Location Sensor Reference
Proximity Gym
RFID [10]
Environment Depth camera [30]
Environment Depth camera [14]
Environment Depth camera [15]
Environment Depth camera [16]
Environment Depth camera [17]
Environment Camera [13]
Gym mat Pressure sensor [12]
Wearable Gloves Pressure sensor [5]
Waist Accelerometer,
gyroscope sensor
Waist Accelerometer,
gyroscope sensor,
Barometric pressure
Ankle, waist Accelerometer,
gyroscope sensor
Smart textile Pressure sensor, EMG [28]
Bicep EMG [29]
Necklace Heart rate sensor [32]
Ear Heart rate sensor [26]
Ear Heart rate sensor [33]
Leg, chest,
and hand
gyroscope sensor
Leg, hand,
and arm
gyroscope sensor
Table 1. Some prominent examples of the
proximity and wearable sensors used for
indoor activity tracking.
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
MyFitnessPal (
Gamification and Exergaming
Gamification [45] and exergaming [46] are two motivation-
al approaches with the goal of changing users’ sedentary
lifestyles. Gamification is defined as the use of video gam-
ing elements in nongame contexts [45]. For example, Con-
solvo et al. [47] developed UbiFit Garden, which embeds
body sensors and a mobile display to encourage their users
to participate in physical activ ities. In addition to gamifi-
cation, another game-related area is exergaming, which is
popular in indoor environments. Oh and Yang [46] defined
the most common exergames as video games that require
physical activity in order to play. In a prominent example,
Nurkkala et al. [48] used video projectors to develop an
exergaming simulator for gym training, exercise testing,
and rehabilitation.
Rabbi et al. [49] and Rawassizadeh et al. [50] found that
users are not motivated to ma nually log their exercise,
especially when they are not accompanied by personal
trainers. Therefore, external motivation, like gamifica-
tion, is required to encourage manual data entry. With
that in mind, Rabbi et al. [49] proposed a vir tual exercise
assistant with a virtual reality head mounted display and
an attached, miniature IoT sensing device that can track
users’ activities. Another prominent example of physical
activ ity gamification is Pokémon GO [51], which resulted
in US$100 million sales a month after its release [90],
although this app does not have an indoor context.
Interaction Interface
Multiple ty pes of interaction interfaces and methods are
used to communicate with users and provide analysis of
collected physical activity information. Through a review
of the literature, we found that GUIs, sound, vibrotactile
sensations, and ambient interfaces are being used for this
purpose. Figure 2 presents an example that includes three
interaction interfaces (except ambient interface).
Modern smartphones are powerful enough to handle
intensive computational tasks such as image processing
[52]. Data gathered from wearable sensors can be trans-
mitted to a smartphone for further processing or visualiza-
tion [53]. However, unlike wearables, smartphones are not
compact enough to be comfortably carried by the user
while exercising [54]. In addition, the proximity of wear-
able devices to users is higher than that of smartphones
[55], which makes them more accurate in collecting users’
physical activity data. Therefore, a hybrid approach that
uses wearables for data collection and smartphones for
analysis and review is common.
Wrist-worn wearables have miniature screens that
present information and collect user input. The most
accepted approach, used by the vast majority of mobile
and wearable applications, is to present physical activities
using numbers and graphs. However, existing approaches
that present information with graphs and numbers lack
actual coaching advice on the exercise sessions [56]. For
example, GUIs may present the calories burned and the
average heart rate during the exercise session, but not
instruct users to adjust body movements and advise them
about how to avoid injury while exercising. This is because
most wearables are attached to users; without a third-per-
son point of view [53], such devices can collect informa-
tion about only a very limited number of body movements.
The screens of wearables and sma rtphones are person-
alized and offer efficient communication, but they still
require users to look at them, which increases users’
cognitive load and shifts their focus [57]. A big screen
mounted on a wall or a training machine can make the
information easier to digest [30] and mitigate the burden of
cognitive load involved in processing information from a
smartphone or wearable. (These ambient interfaces will
be discussed further in later sections.)
The ear is like a biological equivalent of a USB port [3].
Many users listen to music while performing physical exer-
cise [58]. Notably, a study by Nakamura et al. [59] ha s
shown that listening to one’s preferred music can improve
exercise endurance. The widespread use of headphones
while exercising leads to the introduction of headphones
with sensors, i.e., earpieces.
Vibrotactile Sensation
Figure 2. The different types of interaction interfaces.
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
Integrated sensors in earpieces for activity tracking
are growing. Poh et al. [33] introduced a pair of earbuds
that are embedded with ref lective photosensors [often
called photoplethysmography (PPG)] to measure the
user’s heart rate. After connecting the earbuds to the
phone, users can get a ref lection of whether their heart
rate is within the ideal range. A recent proposal from
Matsumoto et al. [60] reported the use of an earpiece
with an eardrum (tympanic) temperature sensor and a
galvanic skin conductance sensor. That earpiece can
monitor vital parameters such as body temperature and
sweat concentration. Advances in computing-capable ear-
pieces have led to their recent adoption by the consumer
market (
Without the distraction from a screen, audio coaching
earbuds use sound to communicate with the user to pro-
vide information and coaching tips such as average heart
rate, pace, time of exercise, and recommendations to
increase or decrease training intensity. The salient advan-
tage of earpieces over GUI is the reduction in cognitive
load so that they can dedicate more of their attention
directly to exercise.
Vibrotactile Sensation
Vibrotactile or tactile sensation is another way to pro-
vide information to and communicate with users. To
our knowledge, existing methods are largely imple-
mented in smart textiles [61]. By providing force feed-
back to certain muscle groups, users can receive
direct guidance for body movements. Hadi et al. [3]
found that haptic feedback improves users’ perfor-
mance in physical activities. In one such application,
Kishishita et al. [62] proposed a force-feedback suit
to provide real-time vibrotactile feedback to the
upper limb with a set of pneumatic gel muscles.
Coaching advice on muscle movement can then be
directly provided to the relevant muscle group. How-
ever, the pressure on different parts of the body could
be confusing [63] for the user and this might lead to
misundersta nding of the given coaching advice. Nev-
ertheless, in comparison to other interfaces, this
force-feedback system places a significantly smaller
cognitive load on the user.
Ambient Interfaces
Gross [64] explained that “ambient interfaces go beyond
the classical GUI and use the whole environment of the
user for the interaction between the user and the system.”
According to Ramos et al. [65], the environment that is
equipped with an ambient interface should recognize the
person who is using the device. The ambient interface
should also provide real-time feedback ba sed on the user’s
needs and behaviors [66].
A prominent example of an ambient interface can be
found in the work of Vales-Alonso et al. [67]. They provid-
ed an ambient interface that personalizes training ses-
sions using a wireless sensor network (WSN) deployed in
the training field. The WSN har vests location-related
data including wind speed, temperature, and so on and
then transmits them to users’ sma rtphones. Users’ train-
ing equipment is equipped with sensors that collect
information such as users’ heart rates. The decision
component located in the training field estimates each
user’s future heart rate and recommends actions based
on environmental conditions (e.g., temperature of the
training field) and the user’s current heart rate. Table 2
provides a comparison to the communication capacity,
obtrusiveness, and personalization among different
interaction interfaces.
Challenges and Opportunities
Despite promising applications for indoor activities, there
are limitations of existing approaches that present oppor-
tunities to further advance the field.
Accuracy and Flexibility
After digital pedometers, smartphones were the first
device used for activ ity recognition. However, due to their
proximity, which did not continuously access the user’s
body movement 24/7 [68], they were replaced by wearable
devices. However, wearables also suffer from inaccuracy
in their data collection [69]).
Khurana et al. [14] found that the detection of ascend-
ing motion, such as climbing stairs, is more accurate when
an iner tial measurement unit is attached to users’ bags
than when it is attached to their shirts. For similar rea-
sons, data collected by smar twatches can not be accurate-
ly used for exercise tracking because it is limited to wrist
movements and thus cannot recognize weight training
moves such as leg presses. Smartwatches are also located
on the wrist, where PPG signals are unreliable due to low
blood flow [70], resulting an inaccurate heart rate values.
The tight connection between a smartwatch and wrist
adds extra pressure to skin that leads to reduced blood
Capability Obtrusiveness
GUI High High High
Sound High Medium High
Low Low High
High Low Low
Table 2. Three different types of
interaction interfaces; the ambient
interface is usually a public display
that is shared among users and not
shown in this figure.
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
flow [71]. Therefore, the heart rate data collected by smart-
watches is not reliable.
To our knowledge, there is a very limited number of
tools and systems available on the market that can track
both aerobic and anaerobic exercise. Currently, fitness
tracking applications on wearable dev ices [e.g., Google Fit
Workout on Google WearOS (https://www.wearableinear
.com/best-heart-rate-monitor-earbuds) and Fitplan on
Apple WatchOS (
-gym-home-workouts/id1064119547) require users to man-
ually input data on anaerobic exercise. This is mainly
because existing approaches can only access a limited
number of body parts. As a solution, Rawassizadeh et al.
[53] proposed the concept of a companion robot that uses
cameras and image recognition algorithms as a third-per-
son perspective to unobtrusively record indoor exercises.
However, no such prototype has been implemented and
evaluated in a real-world setting.
Energy Efficiency
Energy efficiency can typically be achieved by either
advancing battery technology or minimizing energy con-
sumption algorithmically [55]. The size, shape, and capaci-
ty of the battery must be compatible with the compact
design of wearables [9]. Carbon nanomaterial-based (car-
bon nanotubes, graphene, and so forth) batteries are
expected to offer higher energy density and capacity [73],
but they remain under development.
Lithium-ion (Li-ion) and Li-ion polymer batteries are the
most common types used in wearable devices. Due to their
low energy density and shor t life cycle, most wearables
suffer from the need for frequent charging and battery
aging problems [55]. Recently, wireless charging, which
provides more flexibility than cable charging, has become
more popular. Some manufacturers are starting to make
their smartphones with wireless charging capabilities.
Although it can extend the use time of wearables, wireless
charging is still limited by its low speed [74].
Another promising direction is energy harvesting
[75]. Wearables can collect energy from the ambient
environment, such as sunlight [76], [77]; heat [78]; a nd
human motion [79]–[81]. Although a promising field,
energy harvesting from the ambient environment is still
in its infancy.
Security and Privacy
The majority of fitness tracking approaches use Bluetooth
Low Energy (BLE) for wireless communication [55]. How-
ever, studies have shown that BLE is vulnerable to cyber-
attacks such as eavesdropping [82]–[84] and traffic
analysis [82]. Rahman et al. [85] developed a prototype that
can attack Fitbit devices using eavesdropping, injection,
and denial of service to prove the vulnerability of the
device. This raises user privacy concerns, as fitness track-
ing approaches include biometric information. This
becomes more challenging when it is taken consideration
that recent advances are using wearable biometric finger-
prints for digital authentication [86].
Video cameras are not yet in widespread use for fitness
tracking, but they are starting to attract the research com-
munity’s attention [14], [16], [17]. The privacy risks of video
cameras remain, and they are higher than those of other
activ ity tracking devices [87]. More privacy issues arise
when considering a semipublic display, such as a wall-
mounted display in a gym, as some might not want to
share their data on a large screen.
Personalized Professional Intervention
Most coaching approaches implemented for wearables are
not designed to connect users with professionals such as
physicians and personal trainers [88]. Each coaching plan
should be unique because users’ fitness needs vary and a
one-size-fits-all approach does not work. Even common
goals, such as achieving 10,000 steps per day, are not sup-
ported by scientific evidence. To satisfy the need for per-
sonalization, Dharia et al. [89] proposed a system that can
customize each user’s fitness session based on their health
profile and daily activities.
However, without inter vention from a professional and
motivation provided by coaching, users might discontinue
their fitness routine or suffer long-term injur ies. We identi-
fied a scientific gap that will require the research commu-
nity to develop new approaches that provide personalized
professional support to meet the needs of each user.
Although there are applications that connect real profes-
sional coaches to users [e.g., Keep Trainer (https://play
.intl&hl=en_US) and Trainerize (
US)], no machine intelligence is involved. Most coaches
and physicians are still using their human-acquired knowl-
edge to manually evaluate athletes’ performance and phys-
ical condition. This is helpful, but it could be further
augmented by the assistance of AIoT [40] or companion/
coaching robots [72].
There is a growing need for indoor fitness activities. Dur-
ing the COVID-19 pandemic, most people have had to stay
at home to maintain socia l distance. To avoid a sedentary
lifestyle, more people have realized the importance of
seeking coaching advice and finding motivation to keep
their indoor fitness routine on track. In this article, we
first reviewed existing data collection tools and classified
them according to the proximity to the users. Next, we dis-
cussed various motivational approaches to help users stay
committed to nonsedentary behaviors, including coaching,
socialization, gamification, and exergaming. We then
looked at the interaction interface of these approaches.
Finally, we highlighted the challenges and future opportu-
nities for both the research community and the commer-
cial market.
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
We thank Jian Cao for drawing the illustrations used in
this article.
About the Authors
Tianyi Wang ( earned his B.A. degree
in finance in 2015 from Kent State University, his M.Sc.
degree in financial management in 2018, and his M.Sc.
degree in computer information systems in 2020 from
Boston University. His research interests include artifi-
cial intelligence, natural language processing, machine
learning, ubiquitous computing, and wearables.
Yanglei Gan ( earned his B.A.
degree in economics and mathematics in 2018 from the
University of Connecticut and his M.Sc. degree in com-
puter information systems in 2019 from Boston Universi-
ty, where he is currently a graduate student in the Depart-
ment of Computer Science, Metropolitan College, Boston,
Massachusetts, USA. His research interests include artifi-
cial intelligence, natural language processing, machine
learning, ubiquitous computing, and wearables.
Scott D. Arena ( is a senior lecturer in
the Department of Computer Science at Metropolitan
(MET) College, Boston University, Massachusetts, USA. He
spent 38 years in the R&D labs of Bell, where he worked for
ATT/Bell/Verizon Labs. He has been granted nearly a dozen
patents in the areas of file optics/networking and security.
He joined MET full time and is the networking curriculum
coordinator, where he enjoys high-order networking and
security-related coursework.
Lubomir T. Chitkushev ( earned his Dipl.
Ing. degree in electrical engineering from the University of
Belgrade, his M.Sc. degree in biomedical engineering from
the Medical College of Virginia at Virginia Commonwealth
University, and his Ph.D. degree in biomedical engineering
from Boston University, where he is an associate professor
of computer science and the director of health informatics
and health sciences at Metropolitan College, Boston Uni-
versity, Boston, Massachusetts, USA. He is also an associ-
ate director of the Center for Reliable Information Systems
and Cyber Security at Boston University. His research inter-
ests are focused on the modeling of complex systems, com-
puter networks security and architecture, and biomedical
and health informatics.
Guanglan Zhang ( earned her
Ph.D. degree from the School of Computer Engineering,
Nanyang Technological University, Singapore, for her
doctoral work in bioinformatics. She is currently an
associate professor in the Department of Computer Sci-
ence, Metropolitan College, Boston University, Boston,
Massachusetts, USA. Her research interests include the
computational modeling of complex biological process-
es, such as the identification of vaccine targets; develop-
ing solutions for biomedical data management; and the
building of analytical tools for pattern recognition from
biomedical data.
Reza Rawassizadeh ( earned his B.Sc.
degree in software engineering and his M.Sc. and Ph.D.
degrees in computer science from the University of Vienna,
Austria. He is currently an assistant professor in the Depart-
ment of Computer Science and is affiliated faculty in the
Health Informatics Research Lab, Metropolitan College,
Boston University, Boston, Massachusetts, USA. His
research interests include on-device machine learning, ubiq-
uitous computing, wearables, and interaction interfaces for
fitness activity tracking.
[1] N. Peirce, “Diabetes and exercise,” British J. Sports Med., vol. 33, no. 3, pp.
161172, 1999. doi: 10.1136/bjsm.33.3.161.
[2] M. Verissimo et al., “Physical exercise and thrombotic risk in the elderly,” Portu-
guese J. Cardiol., vol. 20, no. 6, pp. 625639, 2001.
[3] S. A. Swaid, N. M. Khanfar, and D. Loudon, “Developing a competitive marketing strat-
egy: The case of a complete fitness gym,” Competit. Forum, vol. 17, no. 2, pp. 258266, 2019.
[4] W. R . Thompson, “Worldwide survey of fitness trends for 2020,” ACSM’s Health Fit-
ness J., vol. 23, no. 6, pp. 1018, 2019.
[5] A. Akpa, M. Fujiwara, H. Suwa, Y. Arakawa, and K. Yasumoto, “A smart glove
to track fitness exercises by reading hand palm,” J. Sensors, vol. 2019, 2019. doi:
[6] K.-H. Chang, M. Y. Chen, and J. Canny, “Tracking free-weight exercises,” in Inter-
national Conference on Ubiquitous Computing, J. Krumm, G. D. Abowd, A. Senevi-
ratne, and T. Strang, Eds. Berlin: Springer-Verlag, 2007, pp. 1937.
[7] R. Rawassizadeh, “A holistic multi-purpose life logging framework,” Ph.D. disserta-
tion, Univ. Wien, Austria, 2012.
[8] M. Cornacchia, K. Ozcan, Y. Zheng, and S. Velipasalar, “A survey on activity detec-
tion and classification using wearable sensors,” IEEE Sensors J., vol. 17, no. 2, pp.
386403, 2016. doi: 10.1109/JSEN.2016.2628346.
[9] M. Homayounfar et al., “Understanding smartwatch battery utilization in the wild,”
Sensors, vol. 20, no. 13, p. 3784, 2020. doi: 10.3390/s20133784.
[10] R. Chaudhri, J. Lester, and G. Borriello, “An RFID based system for monitoring
free weight exercises,” in Proc. 6th ACM Conf. Embedded Netw. Sensor Syst., 2008, pp.
431432. doi: 10.1145/1460412.1460488.
[11] H. Ding et al., “FEMO: A platform for free-weight exercise monitoring with
RFIDs,” in Proc. 13th ACM Conf. Embedded Netw. Sensor Syst., 2015, pp. 141154. doi:
[12] M. Sundholm, J. Cheng, B. Zhou, A. Sethi, and P. Lukowicz, “Smart-mat: Recogniz-
ing and counting gym exercises with low-cost resistive pressure sensing matrix,” in
Proc. ACM Int. Joint Conf. Pervasive and Ubiquitous Comput., 2014, pp. 373382. doi:
[13] A. Nagarkoti, R. Teotia, A. K. Mahale, and P. K . Das, “Realtime indoor workout analy-
sis using machine learning & computer vision,” in Proc. 41st Annu. Int. Conf. IEEE Eng.
Med. Biol. Soc. (EMBC), 2019, pp. 14401443. doi: 10.1109/EMBC.2019.8856547.
[14] R. Khurana, K. Ahuja, Z. Yu, J. Mankoff, C. Harrison, and M. Goel, “Gymcam:
Detecting, recognizing and tracking simultaneous exercises in unconstrained scenes,”
in Proc. ACM Interactive, Mobile, Wearable Ubiquitous Technol., vol. 2, no. 4, pp. 1–17,
2018. doi: 10.1145/3287063.
[15] B.-S. Lin, L.-Y. Wang, Y.-T. Hwang, P.-Y. Chiang, and W.-J. Chou, “Depth-camera-based
system for estimating energy expenditure of physical activities in gyms,” IEEE J. Biomed.
Health Inform., vol. 23, no. 3, pp. 10861095, 2018. doi: 10.1109/JBHI.2018.2840834.
[16] B. Reily, H. Zhang, and W. Hoff, “Real-time gymnast detection and performance
analysis with a portable 3d camera,” Comput. Vision Image Understanding, vol. 159,
pp. 154163, 2017. doi: 10.1016/j.cviu.2016.11.006.
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
[17] G. Barry, P. Van Schaik, A. MacSween, J. Dixon, and D. Martin, “Exergaming
(XBOX Kinect™) versus traditional gym-based exercise for postural control, flow and
technology acceptance in healthy adults: A randomised controlled trial,” BMC Sports
Sci., Med. Rehabil., vol. 8, no. 1, p. 25, 2016. doi: 10.1186/s13102-016-0050-0.
[18] P. Klasnja and W. Pratt, “Healthcare in the pocket: Mapping the space of mobile-
phone health interventions,” J. Biomed. Informat., vol. 45, no. 1, pp. 184198, 2012.
doi: 10.1016/j.jbi.2011.08.017.
[19] O. D. Incel, M. Kose, and C. Ersoy, “A review and taxonomy of activity recognition on
mobile phones,” BioNanoSci., vol. 3, no. 2, pp. 145171, 2013. doi: 10.1007/s12668-013-
[20] R. Rawassizadeh, M. Tomitsch, K. Wac, and A. M. Tjoa, “UbiqLog: A generic mobile
phone-based life-log framework,” Personal Ubiquitous Comput., vol. 17, no. 4, pp.
621637, 2013. doi: 10.1007/s00779-012-0511-8.
[21] G. Fortino, R. Giannantonio, R. Gravina, P. Kuryloski, and R. Jafari, “Enabling
effective programming and flexible management of efficient body sensor network
applications,” IEEE Trans. Human-Mach. Syst., vol. 43, no. 1, pp. 115133, 2012. doi:
[22] G. Fortino, S. Galzarano, R. Gravina, and W. Li, “A framework for collaborative
computing and multi-sensor data fusion in body sensor networks,” Inform. Fusion, vol.
22, pp. 5070, Mar. 2015. doi: 10.1016/j.inffus.2014.03.005.
[23] M. Zheng, P. X. Liu, R. Gravina, and G. Fortino, “An emerging wearable world:
New gadgetry produces a rising tide of changes and challenges,” IEEE Syst.,
Man, and Cybern. Mag., vol. 4, no. 4, pp. 6–14, 2018. doi: 10.1109/MSMC.2018.
[24] A. C. Jimenez-Moreno et al., “Analyzing walking speeds with ankle and wrist worn
accelerometers in a cohort with myotonic dystrophy,” Disabil. Rehabil., vol. 41, no. 24,
pp. 29722978, 2019. doi: 10.1080/09638288.2018.1482376.
[25] Y. Nam and J. W. Park, “Child activity recognition based on cooperative fusion
model of a triaxial accelerometer and a barometric pressure sensor,” IEEE J. Biomed.
Health Inform., vol. 17, no. 2, pp. 420426, 2013.
[26] D. K. Cruz, G. de Graaf, J. Haartsen, F. Hooijschuur, and P. French, “Heart rate
monitoring using infrared thermometry in an earpiece,” in Proc. IEEE Sensors 2017,
pp. 1–3. doi: 10.1109/ICSENS.2017.8234397.
[27] B. Zhou, G. Bahle, L. Fürg, M. S. Singh, H. Z. Cruz, and P. Lukowicz, “Trainwear: A
real-time assisted training feedback system with fabric wearable sensors,” in Proc. IEEE
Int. Conf. Pervasive Comput. and Commun. Workshops (PerCom Workshops), 2017, pp.
[28] B. Zhou, M. Sundholm, J. Cheng, H. Cruz, and P. Lukowicz, “Never skip leg day:
A novel wearable approach to monitoring gym leg exercises,” in Proc. IEEE Int. Conf.
Pervasive Comput. and Commun. (PerCom), 2016, pp. 1–9.
[29] Z. Taha, C. M. Lee, N. U. Ahamed, S. Joseph, and S. F. S. Omar, “Performance
analysis in strength training: An innovative instrumentation,” Procedia Eng., vol. 147,
pp. 455460, 2016. doi: 10.1016/j.proeng.2016.06.340.
[30] A. A. E. Hippocrate, E. T. Luhanga, T. Masashi, K. Watanabe, and K. Yasumoto,
Smart gyms need smart mirrors: Design of a smart gym concept through contextual
inquiry,” in Proc. ACM Int. Joint Conf. Pervasive and Ubiquitous Comput. and Proc.
ACM Int. Symp. Wearable Comput., 2017, pp. 658661.
[31] Y. Hikihara, C. Tanaka, Y. Oshima, K. Ohkawara, K. Ishikawa-Takata, and S.
Tanaka, “Prediction models discriminating between nonlocomotive and locomotive
activities in children using a triaxial accelerometer with a gravity-removal physical
activity classification algorithm,” PloS One, vol. 9, no. 4, p. e94940, 2014. doi: 10.1371/
[32] A. A. Iskandar, R. Kolla, K. Schilling, and W. Voelker, “A wearable 1-lead necklace
ECG for continuous heart rate monitoring,” in Proc. IEEE 18th Int. Conf. e-Health
Netw., Appl. and Services (Healthcom), 2016, pp. 1–4. doi: 10.1109/HealthCom.2016.
[33] M.-Z. Poh, K. Kim, A. Goessling, N. Swenson, and R. Picard, “Cardiovascular moni-
toring using earphones and a mobile device,” IEEE Pervasive Comput., vol. 11, no. 4,
pp. 1826, 2010. doi: 10.1109/MPRV.2010.91.
[34] C. Seeger, A. Buchmann, and K. Van Laerhoven, “myHealthAssistant: A phone-
based body sensor network that captures the wearer’s exercises throughout the day,”
in Proc. 6th Int. Conf. Body Area Networks, 2011, pp. 1–7.
[35] D. Morris, T. S. Saponas, A. Guillory, and I. Kelner, “RecoFit: Using a wearable sen-
sor to find, recognize, and count repetitive exercises,” in Proc. SIGCHI Conf. Human
Factors Comput. Syst., 2014, pp. 32253234. doi: 10.1145/2556288.2557116.
[36] H. Hiilloskorpi, M. Pasanen, M. Fogelholm, R. M. Laukkanen, and A. Mänttäri,
Use of heart rate to predict energy expenditure from low to high activity levels,” Int.
J. Sports Med., vol. 24, no. 5, pp. 332336, 2003. doi: 10.1055/s-2003-40701.
[37] R. Rawassizadeh et al., “Energy-efficient integration of continuous context sens-
ing and prediction into smartwatches,” Sensors, vol. 15, no. 9, pp. 22,61622,645, 2015.
doi: 10.3390/s150922616.
[38] T. De Pessemier and L. Martens, “Heart rate monitoring, activity recognition,
and recommendation for e-coaching,” Multimedia Tools Appl., vol. 77, no. 18, pp.
23,31723,334, 2018. doi: 10.1007/s11042-018-5640-2.
[39] J. Li, H. Cui, T. Guo, Q. Hu, and Y. Shen, “Efficient fitness action analysis based on
spatio-temporal feature encoding,” in Proc. IEEE Int. Conf. Multimedia & Expo Work-
shops (ICMEW), 2020, pp. 1–6. doi: 10.1109/ICMEW46912.2020.9106049.
[40] W. C.-C. Chu, C. Shih, W.-Y. Chou, S. I. Ahamed, and P.-A. Hsiung, “Artificial intel-
ligence of things in sports science: Weight training as an example,” Computer, vol. 52,
no. 11, pp. 5261, 2019. doi: 10.1109/MC.2019.2933772.
[41] N. D. Lane, Y. Xu, H. Lu, A. T. Campbell, T. Choudhury, and S. B. Eisenman,
Exploiting social networks for large-scale human behavior modeling,” IEEE Pervasive
Comput., vol. 10, no. 4, pp. 4553, 2011. doi: 10.1109/MPRV.2011.70.
[42] X. Gui, Y. Chen, C. Caldeira, D. Xiao, and Y. Chen, “When fitness meets social net-
works: Investigating fitness tracking and social practices on werun,” in Proc. CHI Conf.
Human Factors Comput. Syst., 2017, pp. 16471659. doi: 10.1145/3025453.3025654.
[43] J. Häkkilä, L. Ventä-Olkkonen, H. Shi, V. Karvonen, Y. He, and M. Häyrynen, “Jog-
ging in a virtual city,” in Proc. 12th Int. Conf. Mobile and Ubiquitous Multimedia,
2013, pp. 1–2. doi: 10.1145/2541831.2541881.
[44] F. Cassola et al., “Online-gym: Multiuser virtual gymnasium using rinions and
multiple kinect devices,” in Proc. 6th Int. Conf. Games and Virtual Worlds for Serious
Appl. (VS-GAMES), 2014, pp. 1–4. doi: 10.1109/VS-Games.2014.7012164.
[45] S. Deterding, D. Dixon, R. Khaled, and L. Nacke, “From game design elements
to gamefulness: Defining gamification,” in Proc. 15th Int. Academic MindTrek Conf.,
Envisioning Future Media Environ., 2011, pp. 9–15. doi: 10.1145/2181037.2181040.
[46] Y. Oh and S. Yang, “Defining exergames & exergaming,” in Proc. Meaningful Play,
2010, pp. 1–17.
[47] S. Consolvo et al., “Activity sensing in the wild: A field trial of ubifit garden,” in
Proc. SIGCHI Conf. Human Factors Comput. Syst., 2008, pp. 17971806.
[48] V.-M. Nurkkala, J. Kalermo, and T. Jarvilehto, “Development of exergaming simu-
lator for gym training, exercise testing and rehabilitation,” J. Commun. Comput., vol.
11, pp. 403411, 2014. doi: 10.17265/1548-7709/2014.05 001.
[49] F. Rabbi, T. Park, B. Fang, M. Zhang, and Y. Lee, “When virtual reality meets inter-
net of things in the gym: Enabling immersive interactive machine exercises,” in Proc.
ACM Interact., Mobile, Wearable Ubiquitous Technol., vol. 2, no. 2, pp. 1–21, 2018. doi:
[50] R. Rawassizadeh, E. Momeni, C. Dobbins, P. Mirza-Babaei, and R. Rahnamoun, “Lesson
learned from collecting quantified self information via mobile and wearable devices,” J.
Sensor Actuator Netw., vol. 4, no. 4, pp. 315335, 2015. doi: 10.3390/jsan4040315.
[51] T. Althoff, R. W. White, and E. Horvitz, “Influence of Pokémon go on physical activ-
ity: Study and implications,” J. Med. Internet Res., vol. 18, no. 12, p. e315, 2016. doi:
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
[52] R. Rajendran and J. Rajendiran, “Image analysis using smartphones for medical
applications: A survey,” in Intelligent Pervasive Computing Systems for Smarter
Healthcare, A. K. Sangaiah, S. Shantharajah, P. Theagarajan, Eds. Hoboken, NJ:
Wiley, 2019, pp. 275–290.
[53] R. Rawassizadeh, T. Sen, S. J. Kim, C. Meurisch, H. Keshavarz, M. Mühlhäuser, and
M. Pazzani, “Manifestation of virtual assistants and robots into daily life: Vision and
challenges,” CCF Trans. Pervasive Comput. Interact., vol. 1, no. 3, pp. 1–12, 2019. doi:
[54] R. Rawassizadeh, B. A. Price, and M. Petre, “Wearables: Has the age of smartwatches
finally arrived?Commun. ACM, vol. 58, no. 1, pp. 4547, 2014. doi: 10.1145/2629633.
[55] S. Seneviratne et al., “A survey of wearable devices and challenges,” IEEE
Commun. Surv. Tuts., vol. 19, no. 4, pp. 2573–2620, 2017. doi: 10.1109/COMST.2017.
[56] T. Kari and P. Rinne, “Influence of digital coaching on physical activity: Motivation
and behaviour of physically inactive individuals,” in Bled eConference. Univ. of Maribor
Press, Bled, Slovenia, 2018.
[57] H. Jeong, H. Kim, R. Kim, U. Lee, and Y. Jeong, “Smartwatch wearing behavior
analysis: A longitudinal study,” in Proc. ACM Interact., Mobile, Wearable Ubiquitous
Technol., vol. 1, no. 3, pp. 1–31, 2017. doi: 10.1145/3131892.
[58] C. I. Karageorghis, Applying Music in Exercise and Sport. USA: Human Kinetics,
[59] P. M . Nakamura, G. Pereira, C. B. Papini, F. Y. Nakamura, and E. Kokubun,
Effects of preferred and nonpreferred music on continuous cycling exercise
performance,” Percept. Motor Skills, vol. 110, no. 1, pp. 257264, 2010. doi: 10.2466/
[60] K. Matsumoto, Y. Temiz, H. Taghavi, E. L. Cornelius, H. Mori, and B. Michel, “An
earbud-type wearable (a hearable) with vital parameter sensors for early detection
and prevention of heat-stroke,” in Proc. 41st Annu. Int. Conf. IEEE Eng. Med. Biol.
Soc. (EMBC), 2019, pp. 70497055.
[61] M. Umair, C. Sas, and M. H. Latif, “Towards affective chronometry: Exploring
smart materials and actuators for real-time representations of changes in arousal,” in
Proc. Design. Interact. Syst. Conf., 2019, pp. 14791494.
[62] Y. Kishishita, S. Das, A. V. Ramirez, C. Thakur, R. Tadayon, and Y. Kurita,
Muscleblazer: Force-feedback suit for immersive experience,” in Proc. IEEE Conf.
Virtual Reality and 3D User Interf. (VR), 2019, pp. 1813–1818. doi: 10.1109/VR.2019.
[63] K. Cherenack and L. van Pieterson, “Smart textiles: Challenges and opportuni-
ties,” J. Appl. Phys., vol. 112, no. 9, p. 091301, 2012. doi: 10.1063/1.4742728.
[64] T. Gross, “Ambient interfaces: Design challenges and recommendations,” in Proc.
10th Int. Conf. Human Comput. Interact.: Theory Pract., 2003, pp. 6872.
[65] C. Ramos, J. C. Augusto, and D. Shapiro, “Ambient intelligence—the next step for artifi-
cial intelligence,” IEEE Intell. Syst., vol. 23, no. 2, pp. 1518, 2008. doi: 10.1109/MIS.2008.19.
[66] E. Aarts and R. Wichert, “Ambient intelligence,” in Technology Guide, H. J. Bullinger,
Ed. Berlin: Springer, 2009, pp. 244249.
[67] J. Vales-Alonso et al., “Ambient intelligence systems for personalized sport train-
ing,” Sensors, vol. 10, no. 3, pp. 23592385, 2010. doi: 10.3390/s100302359.
[68] A. K. Dey, K. Wac, D. Ferreira, K. Tassini, J.-H. Hong, and J. Ramos, “Getting
closer: An empirical investigation of the proximity of user to their smart phones,” in
Proc. 13th Int. Conf. Ubiquitous Comput., 2011, pp. 163172.
[69] R. Rawassizadeh, H. Keshavarz, and M. Pazzani, “Ghost imputation: Accurately
reconstructing missing data of the off period,” IEEE Trans. Knowl. Data Eng., vol. 32,
no. 11, 2019. doi: 10.1109/TKDE.2019.2914653.
[70] S. Lee, H. Shin, and C. Hahm, “Effective PPG sensor placement for reflected
red and green light, and infrared wristband-type photoplethysmography,” in Proc.
18th Int. Conf. Adv. Commun. Technol. (ICACT), 2016, pp. 556558. doi: 10.1109/
[71] A. F. Manso, A. L. Fred, R. C. Neves, and R. C. Ferreira, “Smart-wearables and
heart-rate assessment accuracy,” in Proc. IEEE Int. Symp. Med. Meas. and Appl.
(MeMeA), 2018, pp. 1–5. doi: 10.1109/MeMeA.2018.8438812.
[72] A. Javaheri et al., “Public vs media opinion on robots and their evolution over
recent years,” CCF Trans. Pervasive Comput. Interact., vol. 2, no. 3, pp. 1–17, 2020.
doi: 10.1007/s42486-020-00035-1.
[73] Z. Wu et al., “Carbon-nanomaterial-based flexible batteries for wearable electron-
ics,” Adv. Mater., vol. 31, no. 9, p. 1800716, 2019. doi: 10.1002/adma.201800716.
[74] J. Chen, C. W. Yu, and W. Ouyang, “Efficient wireless charging pad deployment in
wireless rechargeable sensor networks,” IEEE Access, vol. 8, pp. 39,05639,077, Feb.
2020. doi: 10.1109/ACCESS.2020.2975635.
[75] J. A. Paradiso and T. Starner, “Energy scavenging for mobile and wireless
electronics,” IEEE Pervasive Comput., vol. 4, no. 1, pp. 1827, 2005. doi: 10.1109/
[76] K. Lin et al., “Heliomote: Enabling long-lived sensor networks through solar energy
harvesting,” in Proc. 3rd Int. Conf. Embedded Netw. Sensor Syst., 2005, pp. 309309.
[77] V. Raghunathan, A. Kansal, J. Hsu, J. Friedman, and M. Srivastava, “Design con-
siderations for solar energy harvesting wireless embedded systems,” in Proc. 4th Int.
Symp. Inf. Process. Sensor Netw. (IPSN 2005), 2005, pp. 457462.
[78] V. Leonov, T. Torfs, P. Fiorini, and C. Van Hoof, “Thermoelectric converters of
human warmth for self-powered wireless sensor nodes,” IEEE Sensors J., vol. 7, no. 5,
pp. 650657, 2007. doi: 10.1109/JSEN.2007.894917.
[79] P. D . Mitcheson, E. M. Yeatman, G. K. Rao, A. S. Holmes, and T. C. Green, “Energy
harvesting from human and machine motion for wireless electronic devices,” Proc.
IEEE, vol. 96, no. 9, pp. 14571486, 2008. doi: 10.1109/JPROC.2008.927494.
[80] J. Yun, S. N. Patel, M. S. Reynolds, and G. D. Abowd, “Design and performance of
an optimal inertial power harvester for human-powered devices,” IEEE Trans. Mobile
Comput., vol. 10, no. 5, pp. 669683, 2010. doi: 10.1109/TMC.2010.202.
[81] T. Von Buren, P. D. Mitcheson, T. C. Green, E. M. Yeatman, A. S. Holmes, and G.
Troster, “Optimization of inertial micropower generators for human walking motion,”
IEEE Sensors J., vol. 6, no. 1, pp. 2838, 2006. doi: 10.1109/JSEN.2005.853595.
[82] M. Ryan, “Bluetooth: With low energy comes low security,” presented as part of
the 7th {USENIX} Workshop on Offensive Technologies, 2013.
[83] K. Fawaz, K.-H. Kim, and K. G. Shin, “Protecting privacy of {BLE} device users,” in
25th {USENIX} Security Symp. ({USENIX} Security 16), 2016, pp. 12051221.
[84] R. Goyal, N. Dragoni, and A. Spognardi, “Mind the tracker you wear: A security
analysis of wearable health trackers,” in Proc. 31st Annu. ACM Symp. Appl. Comput.,
2016, pp. 131136.
[85] M. Rahman, B. Carbunar, and M. Banik, “Fit and vulnerable: Attacks and defenses
for a health monitoring device,” 2013, arXiv:1304.5672.
[86] S. Mare, R. Rawassizadeh, R. Peterson, and D. Kotz, “Saw: Wristband-based
authentication for desktop computers,” in Proc. ACM Interact., Mobile, Wearable Ubiq-
uitous Technol., vol. 2, no. 3, pp. 1–29, 2018. doi: 10.1145/3264935.
[87] R. Hoyle, R. Templeman, S. Armes, D. Anthony, D. Crandall, and A. Kapadia, “Pri-
vacy behaviors of lifeloggers using wearable cameras,” in Proc. ACM Int. Joint Conf.
Pervasive and Ubiquitous Comput., 2014, pp. 571582. doi: 10.1145/2632048.2632079.
[88] P. Krebs and D. T. Duncan, “Health app use among us mobile phone owners: A
national survey,” JMIR MHealth UHealth, vol. 3, no. 4, p. e101, 2015. doi: 10.2196/
[89] S. Dharia, V. Jain, J. Patel, J. Vora, S. Chawla, and M. Eirinaki, “Pro-fit: A personal-
ized fitness assistant framework,” in Proc. Intl. Conf. Software Eng. and Knowledge
Eng., 2016, pp. 386389. doi: 10.18293/SEKE2016-174.
[90] J. V. Chamary, “Why ‘Pokémon GO’ is the world’s most important game,” Forbes,
Feb. 10, 2018. [Online]. Available:
Authorized licensed use limited to: BOSTON UNIVERSITY. Downloaded on January 22,2021 at 19:18:12 UTC from IEEE Xplore. Restrictions apply.
... [cs.HC] 16 Nov 2022 makes them valuable tools for tracking and improving users' health. The positive impact of SW is realized by assessing users' health through collecting biological, environmental, and behavioral information and quick access to them [6]. ...
Full-text available
Smartwatches (SWs) can continuously and autonomously monitor vital signs, including heart rates and physical activities involving wrist movement. The monitoring capability of SWs has several key health benefits arising from their role in preventive and diagnostic medicine. Current research, however, has not explored many of these opportunities, including longitudinal studies. In our work, we gathered longitudinal data points, e.g., heart rate and physical activity, from various brands of SWs worn by 1,014 users. Our analysis shows three common heart rate patterns during sleep but two common patterns during the day. We find that heart rate and physical activities are higher in summer and the first month of the new year compared to other months. Moreover, physical activities are reduced on weekends compared with weekdays. Interestingly, the highest peak of physical activity is during the evening.
... On the broader sports activity recognition landscape, except for approaches based on pressure mats and well-studied wearable IMUs, many novel sensing systems have been proposed as reviews such as [84,89] showed. In [9], 3 passive capacitive sensors were placed on the body to recognize and count gym exercises. ...
While sports activity recognition is a well studied subject in mobile, wearable and ubiquitous computing, work to date mostly focuses on recognition and counting of specific exercise types. Quality assessment is a much more difficult problem with significantly less published results. In this work, we present Quali-Mat: a method for evaluating the quality of execution (QoE) in exercises using a smart sports mat that can measure the dynamic pressure profiles during full-body, body-weight exercises. As an example, our system not only recognizes that the user is doing push-ups, but also distinguishes 5 subtly different types of push-ups, each of which (according to sports science literature and professional trainers) has a different effect on different muscle groups. We have investigated various machine learning algorithms targeting the specific type of spatio-temporal data produced by the pressure mat system. We demonstrate that computationally efficient, yet effective Conv3D model outperforms more complex state-of-the-art options such as transfer learning from the image domain. The approach is validated through an experiment designed to cover 47 quantifiable variants of 9 basic exercises with 12 participants. Overall, the model can categorize 9 exercises with 98.6% accuracy / 98.6% F1 score, and 47 QoE variants with 67.3% accuracy / 68.1% F1 score. Through extensive discussions with both the experiment results and practical sports considerations, our approach can be used for not only precisely recognizing the type of exercises, but also quantifying the workout quality of execution on a fine time granularity. We also make the Quali-Mat data set available to the community to encourage further research in the area.
... Although exchangeability is useful for verification issues, it is not an important property in the application of NN. By contrast, many NN libraries contain a related function, called the Cross-Correlation (CC) function [22], which is almost the same as convolution operation but cannot flip the kernel, as expressed in Eq. (6). ...
Full-text available
The purpose is to study the applicability of digital and intelligent real-time Image Processing (IP) in fitness motion detection under the environment of the Internet of Things (IoT). Given the absence of real-time training standards and possible workout injury problems during fitness activities, an intelligent fitness real-time IP system based on Deep Learning (DL) is implemented. Specifically, the keyframes of the real-time images are collected from the fitness monitoring video, and the DL algorithm is introduced to analyze the fitness motions. Afterward, the performance of the proposed system is evaluated through simulation. Subsequently, the Noise Reduction (NR) performance of the proposed algorithm is evaluated from the Peak Signal-to-Noise Ratio (PSNR), which remains above 20 dB for seriously noisy images (with a noise density reaching up to 90%). By comparison, the PSNR of the Standard Median Filter (SMF) and Ranked-order Based Adaptive Median Filter (RAMF) algorithms are not higher than 10 dB. Meanwhile, the proposed algorithm outperforms other DL algorithms by over 2.24% with a detection accuracy of 97.80%; the proposed system can adaptively detect the fitness motion, with a transmission delay no larger than 1 s given a maximum of 750 keyframes. Therefore, the proposed DL-based intelligent fitness real-time IP algorithm has strong robustness, high detection accuracy, and excellent real-time image diagnosis and processing effect, thus providing an experimental reference for sports digitalization and intellectualization.
... [709][710][711] Detecting environmental humidity can prevent heatstroke in people who work or live in a high-temperature and highhumidity environment. [712][713][714] The ratio of moisture in the air to the highest amount of moisture at a particular air temperature is defined as relative humidity (RH), which is a crucial factor in quantifying the humidity level. The RH of skin can also be evaluated using environmental RH. ...
Wearable multimodal sensors could enable the continuous, non-invasive, precise monitoring of vital human signals critical for remote health monitoring and telemedicine. Atomically thin materials with intriguing physical characteristics, rich chemistry, and extreme sensitivity to external stimuli are attractive for implementing high-performance wearable sensors. Despite the increased interest and efforts in 2D materials-based wearable sensors, reducing the manufacturing and integration costs while improving the product performance remains challenging. Previous review articles provided good coverage discussing the material and device aspects of 2D materials-based wearable devices. However, few reviews discussed the status quo, prospects, and opportunities for the scalable nanomanufacturing of 2D materials wearable sensors for health monitoring. To fill this gap, we have reviewed the recent advances in 2D materials-based wearable health sensors. We discussed the structure design, fabrication processing, the mechanisms of 2D materials-based wearable health sensors, and their applications for human health monitoring. More significantly, we have provided a systematic discussion of the state-of-the-art and technological gaps for enabling future design and nanomanufacturing of 2D materials wearable health sensors. Finally, we discussed the challenges and opportunities associated with the scalable nanomanufacturing of 2D wearable health sensors. This article is protected by copyright. All rights reserved.
... "Personal Monitoring Device" and "Data Tracking" are two examples. We estimate these themes might be affected by the research trends on wearables, started in 2016 [23], [29]. At that time, methods that can make use of personal monitoring devices in healthcare with blockchain technology are not widely explored. ...
Conference Paper
Full-text available
Blockchain is a technology to enable decentralized collaboration among un-trusted entities. Academia and industry are rushing to uncover its potential for their field of interest. Due to the novelty of the technology and its diverse applications, there are some ambiguities in approaches and trends. In this paper, we analyze the scientific publications and patents from the past five years to identify the trends for blockchain integration with healthcare. For this purpose, we have adopted a quantitative (clustering) and qualitative (theme extraction) approach to discover themes and temporal dynamics in academia and industry. Our results shed light on the potential challenges and vision for future works.
Full-text available
During recent years, the fast proliferation of robots in people’s everyday lives calls for a profound examination of public consensus, which is the ultimate determinant of the future of this industry. This paper investigates text corpora, consisting of posts in Google News, Bing News, and Kickstarter, over an 8-year period and Twitter over a 1-year period, to quantify the public’s and media’s opinion about this emerging technology. The results of our analysis demonstrate that news platforms and the public take an overall positive position on robots. However, there is a deviation between news coverage and Twitter users’ attitudes. Among various robot types, sex robots raise the fiercest debate. Besides, based on our analysis the public and news media conceptualization of robotics has altered over recent years. More specifically, a shift from solely industrial-purpose machines, towards more social, assistive, and multi-purpose gadgets is visible.
Full-text available
Smartwatch battery limitations are one of the biggest hurdles to their acceptability in the consumer market. To our knowledge, despite promising studies analyzing smartwatch battery data, there has been little research that has analyzed the battery usage of a diverse set of smartwatches in a real-world setting. To address this challenge, this paper utilizes a smartwatch dataset collected from 832 real-world users, including different smartwatch brands and geographic locations. First, we employ clustering to identify common patterns of smartwatch battery utilization; second, we introduce a transparent low-parameter convolutional neural network model, which allows us to identify the latent patterns of smartwatch battery utilization. Our model converts the battery consumption rate into a binary classification problem; i.e., low and high consumption. Our model has 85.3% accuracy in predicting high battery discharge events, outperforming other machine learning algorithms that have been used in state-of-the-art research. Besides this, it can be used to extract information from filters of our deep learning model, based on learned filters of the feature extractor, which is impossible for other models. Third, we introduce an indexing method that includes a longitudinal study to quantify smartwatch battery quality changes over time. Our novel findings can assist device manufacturers, vendors and application developers, as well as end-users, to improve smartwatch battery utilization.
Full-text available
The rapid development of wireless power transfer technology brings forth innovative vehicle energy solutions and breakthroughs utilizing wireless sensor networks (WSNs). In most existing schemes, wireless rechargeable sensor networks (WRSNs) are generally equipped with one or more wireless charging vehicles (vehicles) to serve sensor nodes (SNs). These schemes solve the energy issue to some extent; however, due to off-road and speed limitations of vehicles, some SNs still cannot be charged in time, negatively affecting the lifetime of the networks. Our work proposes a new WRSN model equipped with one wireless charging drone (drone) with a constrained flight distance coupled with several wireless charging pads (pads) deployed to charge the drone when it cannot reach the subsequent stop. Our model solves this charging issues effectively and overcomes energy capacity limitations of the drone. Thus, a wireless charging pad deployment problem is formulated, which aims to apply the minimum number of pads so that at least one feasible routing path can be established for the drone to reach every SN in a given WRSN. Four feasible heuristics, three based on graph theory and one on geometry, are proposed for this problem. In addition, a novel drone scheduling algorithm, the shortest multi-hop path algorithm, is developed for the drone to serve charging requests with the assistance of pads. We examine the proposed schemes through extensive simulations. The results compare and demonstrate the effectiveness of the proposed schemes in terms of network density, region size and maximum flight distance.
Full-text available
EXECUTIVE SUMMARY This case is designed for use in business courses to help students learn about the competitive environment and operation of a typical small business organization. The case presents a situation of an actual organization in which Jim Press and Ed Curl (all names are disguised) have developed a concept for a Complete Fitness Gym (CFG) to be opened in a southern metropolitan area. Despite the high failure rate of such facilities in the industry, Press and Curl believe that they can establish CFG as a successful business organization.
Full-text available
Apply It! From this article, the reader should understand the following concepts: • Articulate the differences between a fad and a trend • Use the worldwide trends in commercial, corporate, clinical (including medical fitness), and community health fitness industry to further promote physical activity • Study expert opinions about identified fitness trends for 2020.
Full-text available
Similar to how the smartphone and Internet have significantly changed our daily lives, artificial intelligence (AI) applications have started to profoundly affect our everyday lives as well. Two major products of this relatively recent trend are virtual assistants and home robots. They have similar functional characteristics: both interact with users through conversational agents and attempt to imitate human behavior. Home robots host a virtual assistant and have mechanical capabilities as well. There are many discussions about risks, challenges and the future vision associated with the proliferation of AI at the industrial level. These discussions, however, have not yet widely extended to the user level within the context of daily lives. In this article, we provide a review to discuss the benefits, risks, challenges, open questions and the future vision of using virtual assistants and social robots in daily lives.
The interdisciplinary nature of sports science introduces challenges such as multifaceted data collection, accuracy in knowledge formation, and equipment usability. Artificial intelligence of things (AIoT) technology presents a feasible solution adaptable to different sports. Taking weight training as an example, we apply AIoT technology to these challenges.
Conference Paper
These days there are thousands of workout videos available on the internet. Samsung Health2 [1] provides a dedicated section called programs containing short workout videos for various exercises. The goal is to assist people perform these workouts independently on their own. A common observation is that even people who visit gym regularly find it difficult to perform all steps (body pose alignments) in a workout accurately. Continuously doing an exercise incorrectly may eventually cause severe long term injuries. To help solve this problem and provide assistance in form of a visual feedback while performing a workout, we propose a system to analyze user's body posture during a workout and compare it to a professional's reference workout. We represent human body as a collection of limbs and analyze angle between limb pairs to detect errors and provide corrective action to the user. Our system builds on the latest advancements using deep learning for human body pose estimation. We use techniques for time series data alignment like DTW [2] (Dynamic Time Warping) along with optical flow tracking to synchronize user/reference videos. We are able to detect and locate errors in user's activity (pose) very effectively based on some threshold deviation between the limb angles. The system in future can be extended to be used by physicians to monitor patient's recovery following an injury.