Experimental Investigation of the Usfulness of Bracelet Trackers in

Sports and Health Monitoring

Critical Evaluation of a New Handheld Activity Monitoring Device Class

Hans Weghorn

BW Cooperative State University, Kronenstrasse 53A, 70174 Stuttgart, Germany

Keywords: Ubiquitous Sports Apps, Sports Watches, Health Tracker, Fitness Tracker, Sports Tracker.

Abstract: In professional sports and medicine, the use of electronic devices for activity monitoring and controlled

exercising is commonly established since many decades. Due to miniaturization of computer electronics,

sports and health devices became popular for non-elite sports users during the last twenty years, many of

these in the appearance of watch-like systems, e.g. as running computers or as versatile heart rate monitors.

Technologically based on such devices and since few time, various vendors in the sports and health field

started to offer bracelet-like systems, while making the customers believe that the continuous use of such

devices in daily life can be considered even more fashionable and helpful. This paper compares the new

wristband device generation with the established, well-working sports watches. Significant findings about

the sensor quality together with observation of the enforced Internet-based user handling yield a rather

critical reflection about the usefulness of the this new device class for sports and health activity tracking.

1 INTRODUCTION

The positive impact on human health, which is

generated by regular, not too extensive sports

activities, has been proven in scientific

investigations during the last century repeatedly. For

instance, stroke risks can be considerably diminished

by following physical workout plans for men

(Wannamethee and Shaper, 1992) as well as for

women (Church, 2010). Also various other diseases

can be relieved by weekly body exercises (Law et

al., 1991). Consequently in modern world, scientists

aim at education and motivation for increased

physical activity already with the help of electronic

tools from childhood on (Valentín and Howard,

2013) for overcoming problems like overweight as

early as possible (Colagiuri et al., 2010).

In modern investigations, smartphones often are

used in such concepts and research, because these

devices can easily be programmed with specific

software. On the other hand, the electronic device

market offers a wide variety of tools for non-elite

and popular sports (Fig. 1), often with functionalities

closely related to health support scenarios like

monitoring heart rate (HR) or blood pressure. Such

physiological measures do represent also standard

indicators for planning and tracing training units in

professional sports (Arts and Kuipers, 1994).

Figure 1: Personal sports units: Left hand, an elaborated

triathlon watch and a simplified HR monitoring sports

watch are shown, while right hand an "active" smartphone

with a sport sensor test software is visible.

In particular for endurance sports like running or

cycling, wrist-like watches without or in

combination with RF coupled additional sensors are

well established and broadly in use today for

exploiting fundamental knowledge about physical

training effects (Hoppeler et al., 1985). E.g.,

triathlon represents a kind of sports, which asks for

improvement and supplement even in three parallel

disciplines, i.e. swimming, running and cycling. Fig.

1 shows accordingly a triathlon computer (left-most

device), which looks like a wrist watch, and which

can be connected via RF to additional sensors like

124

Weghorn, H.

Experimental Investigation of the Usfulness of Bracelet Trackers in Sports and Health Monitoring - Critical Evaluation of a New Handheld Activity Monitoring Device Class.

DOI: 10.5220/0006085801240133

In Proceedings of the 4th International Congress on Sport Sciences Research and Technology Support (icSPORTS 2016), pages 124-133

ISBN: 978-989-758-205-9

footpods, turning rate for bicycle wheels and pedals,

and HR chest straps.

Quite commonly, the meanwhile established

handheld wrist devices in Fig. 1 are primarily worn

during sports, while physiological measures are

recorded by additional body sensors. The chest strap

for HR sensing often is sensed being inconvenient or

uncomfortable, a restriction that applies also if

smartphones replace the wrist watch computers by

the help of specialised apps. The idea of improving

the UI with smartphone apps and by that implicitly

also the efficiency of such sports trackers was

investigated already in other work (Weghorn, 2015).

Figure 2: Four bracelets from different vendors are

representing a new generation and style of personal sports

and health trackers.

From the experience with the sports tools and

from the availability of the built-in sensors for

motion, orientation and acceleration from the

smartphone segment, a new device class evolved,

which appears more compact in the shell style of

bracelets (Fig. 2). Technologically, such systems try

to overcome the additional RF-linked sensors by

integrated electronics. Heart-rate rate monitoring is

performed by an optical measurement system, which

is directed to the skin below the tracker bracelet,

acceleration or movement activities are detected by

semiconductor sensors directly inside the device. In

consequence, such tools are feasible for all day use,

and that is exactly, what they are being advertised

for. A much broader customer ship can be attracted

by expanding the use from monitoring and control

during ambitious sports to a general and all-time

tracking of people with interest in their health.

Assuming an average customer without too deep

of knowledge and insight in such technologies, the

typical user faces a broad spectrum of vendors (Fig.

2), who are offering an even broader range of

different bracelet devices. Commonly the price of

such tools are in the order of 100 U$/Euros/GBP,

and there were lots of commercial advertisements

around the recent celebration periods, because this

cost range is well feasible for personal presents.

The here assumed typical user cannot know,

which device out of the many offers is the one,

which best maps his or her own application desires,

neither can this user know anything about the quality

of such new tools, which came up in big mass in

short time. On the other hand, feedback of average

customers is exposed on famous Internet market

places, out of thousands reviews a big part reports

critical user experiences. Reports like (Van Arsdale,

2015) are rather typical, and they unveil that there

might arise serious problems with the reliability of

the new bracelet devices in respect to their

advertised main purpose scope. Such observations

seriously put the measurement quality of the new

devices in to question.

Concerning the UI handling prospective, the tiny

devices appear also interesting for research. For user

inputs the units are typically equipped with just one

mechanical input button, which is in some devices

complemented by screen sensitivity to finger

touches and finger wiping. For information output, a

range is used from simple single LEDs, segmented

LED number displays, OLED displays, and paper

like LCD displays. Each vendor follows at the

moment a unique combination of the hardware UI

possibilities, which are primarily restricted by the

surface of the devices. This may appear also

confusing to a customer, who wants to select an

appropriate unit.

Regarding this overall situation, in the research

here the new bracelet device class should be

investigated with scientific methods systematically.

After putting a given selection of typical devices

(Fig. 2) into operation (software installations +

configuration), a measurement series was intended

for investigation of their precision in movement

tracking and in HR monitoring. For the mutual

verification, the above discussed triathlon system

was used, since its reliability was investigated and

verified in former research already (Weghorn,

2014). As sideline result of the UI handling some

indicative and critical findings about the usability of

the Web-based assistive UIs and about privacy

concern could also be derived here.

2 SETTING UP THE BRACELET

TRACKERS FOR OPERATION

In this investigation, a systematic validation of the

measurement quality was targeted for the four

available bracelets, which are visible in Fig. 2. In

particular the following aspects should be evaluated,

Experimental Investigation of the Usfulness of Bracelet Trackers in Sports and Health Monitoring - Critical Evaluation of a New Handheld

Activity Monitoring Device Class

125

because these reflect the primary use features of the

devices:

Step counting – Precision of movement detection

in walking and running.

HR monitoring – Accuracy of HR tracking,

events and periods of possible signal loss.

As look ahead it shall be stated here already, that

the exploration could not be performed to the full

extend like in former, similar research (Weghorn,

2014). The reason for this was, that the systems do

not grant reasonable or full access to their data

scans; some of them are even are reporting data

despite there is no physiological input available at

all. It shall be furthermore stated here, that for legal

reasons, the findings are described in an anonymous

way, so that they cannot be projected to certain,

particular product vendors.

Direct relevance for the handling viability of the

devices - especially during all-day use - arise from

their own physical dimensions. The weight and also

the size (total volume of core device without

watchstrap) of the given set of units was measured

here and is listed in Table 1.

Table 1: Measures for physical dimensions of the different

tracker devices.

weight

[gram]

size

[cu. cm]

sports

35.0

watches

simple

15.8

smart

sports

113

79.2

phones

full+slim

132

77.3

bracelets

9.24

11.34

7.33

7.80

The most compact bracelet devices do have

either no built-in HR sensor (D) or no own display

(C). Compared to a simple sports watch, the size and

weight differences to the bracelets are not too big,

while a fully equipped system like the triathlon

watch computer T requires its considerably bigger

dimensions for hosting all the electronic and

rechargeable battery with long operational time.

In the beginning of the experiments, the systems

had to be set up for operation. In general, this has to

be performed by installing a communication relay

software on one's personal host computer (similar to

a device driver, but a rather elaborated one with

autonomous Internet communication) and by

registering as user on a Web site of the vendor. For

the latter, entering personal data like passwords and

e-mail addresses is mandatory and cannot be

circumvented.

For acquaintance to the handling of the devices,

arm position tests for proper HR sensing were tested.

In accordance to the recommendations for the

devices in their manuals, positioning and launch of

using the HR monitoring on the display appeared

appropriately efficient.

The bracelet devices perform motion tracing in

walking and running through their acceleration

sensors from the forearm position. Of course this is

only indirectly coupled to any foot stepping activity.

Therefore, a manual step counting was performed

with the devices. One hundred walking steps were

counted, and the device display values were noted

on a paper. While walking, the arms were either

moved in synchronous swings to the walking cycle

or the arms were intentionally damped in their

movement. Several laps were conducted for devices

A, B and D, and the maximum counting errors are

shown in Tab 2. Obviously, the devices register two

counts for each natural walking step with both feet.

Table 2: Values registered by the bracelet devices during

the manual counting experiments with 100 walking steps.

bracelet

synchronous arm swing

167

203

200

damped arm swing

203

150

Complementary, it was tested, whether the

devices are registering any body movements

incorrectly as steps due to their forearm position. As

listed in Tab. 3 a set of typical, gymnastic exercises

was used for this experiment.

Table 3: Test of possibly incorrect step counting in typical

gym workouts (sets of 20 repetitions were conducted).

bracelet

squats

counted

dumbbell front raise

counted

dumbbell lateral raise

counted

arm circling

counted

These observations about motion tracking

suggest, that the bracelets can theoretically be used

for endurance workouts in walking and running, if

icSPORTS 2016 - 4th International Congress on Sport Sciences Research and Technology Support

126

Figure 3: The three devices T, A and B were used in parallel in the first activity experiment with four sets of stemming pull-

ups. The upper part shows as reference the measurement curve for the professional triathlon system T indicating HR at

resting start and at its maximum. The middle and the lower curve show the measurements of bracelets A and B respectively.

These indicate a proper map of the resting phase, but totally wrong slope and peak values otherwise.

arm movement is performed with certain discipline

and care. On the other hand, it is pre-programmed

that an all-day use will generate totally wrong

accumulation counts for so-called steps.

The use procedure for the bracelets has to be

performed in the following functional sequence:

1. The user wears the bracelet; either on manual

UI input or on automatically, the bracelet device

starts detecting, measuring and recording

physiological input data.

2. The user has to synchronize the device data (this

also can happen automatically): Through an RF

connection or a data cable the gateway ("device

driver") on the personal computer extracts the

data records from the bracelet and forwards it to

the Web site of the vendor. That means, all data

is collected in a kind of cloud system and is not

under primary access and ownership of the

bracelet user.

3. The user has to log into the vendor's Web portal

for accessing his/her own data. Due to the

limitations of Web page programming, the UI

there has to be handled in an entangled way for

accessing the information. Unlike announced by

the vendors (with words in sense like "the user

owns his data and has full access") only vague

summaries of activity traces can be displayed

and downloaded. Data plots, which are visible

in the Web UI, are not presented in scientific

style, and can therefore not evaluated in this

way (refer to Fig. 4 to Fig. 6).

The software setup for first operation was started

and performed only for three stand alone devices A,

B and D, because the inconvenience of the handling

prevented a quicker advance already in this very first

stage of this work. Next, the devices were used

outdoor in shorter sequences for learning the proper

handling and the methods of extracting the captured

data afterwards. It turned out, that device D cannot

be used in the required way, because it doesn't allow

any lap control.

Therefore, D excluded itself from the further

extended screening. Also two operational bracelets

are sufficient in the beginning, because during

mutual verification one person can only wear two of

such devices on both arms without mechanical

disorder. On base of this, the first systematic

experimental series had been started.

Experimental Investigation of the Usfulness of Bracelet Trackers in Sports and Health Monitoring - Critical Evaluation of a New Handheld

Activity Monitoring Device Class

127

Figure 4: Reference scan of device T for a running experiment with a set of laps; this was carried out for statistically

evaluating the measurement quality of bracelet device that are used in parallel. During the slow down phases, the devices

were handled for starting and ending the individual laps.

3 ENDURANCE EXPERIMENTS

In the following experiments, the bracelet devices A

and B together with triathlon sports watch T were

used. Where required, T was connected to a HR

chest strap and also to a 3D footpod sensor inside

the running shoe of the experimenter. Despite that

the bracelets come with disclaimer, that the devices

are not indented for this kind of use, one of the first

experiments was a sequence of strength training.

Stability and tracking slope in HR sensing

The active part of the stemming experiment started

after a resting phase at resting pulse as displayed in

Fig. 3. After 30 seconds, the experimenter stood up

(seen from the barometric altitude sensing of T

visible in Fig. 3) and executed four series of arm pull

ups with intermediate breaks. The weight load for

the pull-ups was increased after the second set. Since

all three devices T, A and B were worn in parallel,

the following observations can be derived for this

experiment on base of the reference scan from T:

- The constant, resting pulse was registered

correctly by A and B.

- The maximum pulse value is neither detected at

the proper time moment by A and B, nor is it

correct in its absolute value.

- The curves are extremely distorted, only vague

similarity to the reference is found.

This test yields first obvious indication, that the

bracelet devices cannot follow a steady change of

pulse, but they are able to detect values that are

constant over a longer period only. The display style

for the UI of the bracelets can also be commented as

non-scientific and unprofessional. Furthermore, it

requires a rather winding browsing sequence to get

these graphs displayed on the bracelet UI systems at

all.

Statistics about heart rate tracking accuracy by

conducting series of running laps

For a field evaluation of the tracking accuracy, a

series of running laps was performed. Within an "in

and out cycle" one main lap on a cart track along a

river was coped at an intermediate running speed of

approx. 7 mph. This moving speed should incur an

intermediate HR and also an intermediate foot tread

rate. The starting and stopping of the laps had to be

handled on the bracelets A and B in little complicate

manner due to the limited input controls; therefore,

the precision of the lap capture was in the order of

one to two seconds, sometimes worse as seen from

the lap durations in Fig. 6. The proper lap distance

was tracked in parallel by the river landmarks and by

the GPS display of device T.

In the first systematic series, devices T, A and B

were worn and used. After warming up during

reaching the start of the first testing lap, ten

repetitions were performed by using device A in the

"out" round and using device B in the "in" round of

the lap course (Fig. 4). Handling all three devices in

parallel was not possible in a reasonable way, so

there is no direct comparison between A and B

recordings available for this part.

After transferring the data captures of the used

tracking devices to the host UI systems, the

following turned out:

 The bracelet Web UIs do not release the data

about the laps in a way that it can be efficiently

used for a statistical evaluation. The UIs

display only summaries of information, but no

individual measures on the laps to the required

deepness like it is know from the standalone

PC software for device T, which was used in

former investigations before many times.

 Device A ignored the manual lap control

completely and merged the entire time (e.g. for

the run in Fig. 4) into one overall output

display only.

 As mentioned before, with device D no lap

control is possible at all, and it doesn't have an

HR sensor anyway.

icSPORTS 2016 - 4th International Congress on Sport Sciences Research and Technology Support

128

Figure 5: Just one single lap should be recorded with bracelet A by attaching it only to its regular forearm position during

the green time window. This lap was started and stopped on device A by the defined button presses, but in its evaluation UI,

the record shows the full time of the excursion; even for the walking time to the starting way point a data curve is displayed

despite A was worn inside a side pocket, totally detached from skin. Obviously, the bracelet software blindly invents data

points for phases, where it assumes activity, despite there is absolutely no physiological input.

This first running series has to be therefore

considered as fail, because the bracelet UIs do not

provide the desired data content for enabling any

deeper analysis. As consequence, it was decided to

record only one single lap with device A and extract

this immediately from the device for preserving

more detail information. This method - of course -

invokes a much higher experimental overhead,

because at least five laps should be collected for

obtaining reasonable statistics.

In accordance to this, the next approach was

started with a follow-up experiment: Device T and B

were put into their regular arm position, while

device A was put into a side pocket, clearly detached

from any body or skin contact. The experimenter

walked approx. 3 mins to an appropriate starting

point for further tests and started running for

approximately 17 mins from there. During this, three

laps of 1/2 mile each plus intermediate breaks were

applied. As Fig. 5 shows, bracelet A recorded data

already during the walking time despite it was

neither worn properly, nor started manually. For this

reason, a time gap between the two curves in Fig. 4

arises.

Furthermore, the bracelet A software system

blindly invented data scans despite there was

absolutely no physiological input; this is exposed in

the curves before and after the activation window in

Fig. 5. The HR measures in these phases are - of

course - completely wrong and much too high,

which can be derived from the reference scan on

device system T. Only during the activation window

of the intended lap, the curves between A and T are

aligned to a reasonable degree.

A hardware/software systems, which invents data

randomly and exposes the impression that the

system is working correctly despite it is not, clearly

can be excluded as serious device. This applies

especially, if sensitive data like human HR is to be

detected. Further investigation about HR tracking

with bracelet A can by that be derived being totally

nonsensical, further research was therefore stopped

at this point with unit A.

For bracelet device B, the situation arises much

less erroneous. Although no detail scans about the

HR curves can be extracted in its Web UI, Fig. 5

tabulates the number results for the five laps that

were recorded in this experiment with device B. As

explained before with Fig. 4, 20 laps have been

captured with device T in total, while A and B were

used alternately and in parallel to T. Fig. 5 shows the

result records for T that corresponded to the B using

slots. Overall, the averaged and maximum HR

values for the five laps show, that there is only a

maximum difference of two beats/minute between

the two different systems, while most values are

almost identical or within a difference of just one

count. This implies, that pulse capture with device B

Experimental Investigation of the Usfulness of Bracelet Trackers in Sports and Health Monitoring - Critical Evaluation of a New Handheld

Activity Monitoring Device Class

129

Figure 6: Upper part shows the output table of the Web UI for bracelet device B, which lists a series of five lap traces out of

the full set in Fig. 4. Interestingly, the stride counting is not reported, but the covered distances and derived moving speeds

are shown. With device B, the distance measure systematically is to high, because again all laps covered 0.5 mi as validated

by GPS and landmarks. The lower table is a mosaic from the corresponding running details, which were simultaneously

recorded with device T and were displayed by T's standalone software for personal computers.

seem to work appropriate for slow variations of this

type of measure.

Statistics about motion tracking accuracy

Since the Web UI of device A doesn't allow to

record and display a set of laps individually, a test

on movement tracking was performed by manually

handling this bracelet through its one button input.

In this way, it is possible to read the actual step

counting and total distance on the tiny number

display. A series of five walking laps with 0.5 miles

each was collected (Tab. 2), and the intermediate

counting values of device A were written on a paper.

By mutual verification with T and with the

knowledge from former experiments on this kind of

research (Weghorn, 2014), the step counting was

found to be appropriately precise for this sensor

concept.

Based on an internal conversion factor for the

step size, which cannot be modified, the device

registers systematically too low a covered moving

distance (in average 0.415 mi instead of 0.5 mi) with

some intrinsic error bar, which is caused by the

natural variation of step size during walking. This

means, that with device A the moving distance has

Table 4: Hand written protocol of five movement tracking

laps of 0.5 miles with device A.

lap no.

steps [counts]

753

703

725

730

718

distance [miles]

0.70

0.61

0.67

0.68

0.66

high repetition quality, but shows also a

considerable absolute error, because the lack of any

calibration procedure. The latter is standard for

sports computers like T.

Furthermore, it turned out as side observation in

the test series of Tab. 4, that device A registered

floor levels while moving the arm up and down for

reading the values from the display. This suggests

that the floor level counting is also performed in a

nonsensical way, and its result is worthless.

In contradiction to the situation with device A,

the Web UI of device B lists individual laps out of

one bigger activity as these are launched and

terminated through the bracelet input menu. Fig. 5

shows a set of five lap runs, from which it can be

derived that the covered distance is registered as

systematically too high. In consequence, the moving

speed is also wrong inversely. For unknown reasons,

the Web UI does not show the counting of the foot

steps, so it cannot be validated, whether the distance

error refers to a wrong step length parameter, which

also cannot be calibrated in this device. There is a

time gap of approx. one minute in Fig. 5 between the

recordings of device B and T, but the latter system is

based on GPS and therefore absolutely precise.

In summary, the movement distance tracking can

be considered as comparably precise to the method

based on the footpod sensors despite the movement

is detected by the bracelet on the forearm, which

does not have a 1:1 relation to foot stepping. The

concept suffers anyway from an intrinsic problem,

because in general step width is a varying value and

not a fixed parameter.

icSPORTS 2016 - 4th International Congress on Sport Sciences Research and Technology Support

130

4 CRITICAL DISCUSSION

The before described experimental results have

shown, that the bracelet trackers are not feasible as

single device solution for sports and health

scenarios. Obviously, the established approach of

additional - but somehow inconvenient - sensors like

heart chest straps represents the more accurate and

reliable method. It can be assumed that this is well

realized by the vendors of the device samples B and

D, who enable their devices to be linked with such

additional, professional sensors. Since these firms

develop and offer also systems like T since many

decades, they are experts in this field of

physiological sports sensing.

The replacement of the HR chest strap sensor for

pulse detection by the optical reflection system can

be considered as fail in all applications, where

movement and quicker pulse variations apply. That

also such an optical system suffers from signal

detection delays is known from all finger pulse

sensors, which are in use since a long time in

medicine and in clinical environments (Fig. 7).

Unlike the bracelets, these medical devices optically

shine through the finger for detecting the vascular

pulse contractions, which is a better then the

possibly moving and by that unstable reflection

method of the tracker bracelets. Despite this better

construction the medical sensors typically require

around a minute for synchronizing to the pulse.

The observation that one of the investigated

bracelet tracker system invents blindly HR data in

case of signal loss, appears almost unbelievable.

This behaviour generates the illusion that the system

is working, while it is in reality often incapable for

its intended purpose. This represents a clear fraud by

the responsible engineers and companies in this very

sensitive application field of tracking human

physiological conditions.

There is already research conducted, which tries

to employ such systems in elderly care (Alsulami et

al., 2016), and which needs to rely on the advertised

properties of the bracelet devices in their concepts.

In emergency situations, such sensors could

therefore even lead to disastrous results, but it is also

not new that researchers uncritically trust in the

sensor quality of modern consumer devices

(Valentín et al., 2013).

Regarding motion tracking, the registration of

foot steps appear to work in most cases as precise

and reliable as known from other, comparable

systems (Weghorn, 2014). Unfortunately, there is no

calibration of the foot step size in the tested bracelet

devices, and therefore all come up with considerable

Figure 7: Optical pulse detection is standard in medicine,

where monitors like the pulsoxiometers (right device)

shine an optical beam through the finger. This yields better

contrast than the reflection method of the tested bracelet

trackers. Here again, the obtrusive unit (left device)

enforces measurement of just any signal, which leads to a

totally wrong result of 118 bpm in contradiction to the

correct sensing value of 65 bpm.

measurement errors in tracing distances and

velocities. Calibration of the stride length was

standard from the beginning during the use of

footpods with devices like T. This was introduced

for providing reasonable functionality of such

devices also indoor. Outdoor, such sports computers

like also health and sports apps in smartphones are

using GPS, which is clearly the better localization

technology, and which is currently being introduced

in improved versions of device B. All this will lead

to bracelet constructions, which are constructed

internally like the original computerized sports

watches, while only the shell appearance remains

different at the end of this evolution.

Of course, it is not possible to test the entire

variety of available device on the market. Hence, it

cannot be proven that there do not exist well-

working bracelet devices anywhere. A counter-

productive handling style and automatic mechanisms

like lap data merge prevented the targeted

experimental advance and deepness in this research.

Independent of this, the major point can be

addressed here sufficiently, that non-expert users are

not able to distinguish the functional from the non-

functional bracelets. The current amount of user

feedbacks, which is exposed by the bigger Internet

market places, show that there are, e.g., ten

thousands of owners of device A and a big portion

of them are complaining about the device. A fraction

of them are addressing problems with measurement

and tracking accuracy, but in sum all these owners

are victims of a fraudulent product, although most of

them may not be able to realize this themselves.

Despite all technical restrictions, the bracelet

trackers can also be interpreted of having some

Experimental Investigation of the Usfulness of Bracelet Trackers in Sports and Health Monitoring - Critical Evaluation of a New Handheld

Activity Monitoring Device Class

131

positive effects. Although they may work imprecise

or completely wrong in scientific or medical sense,

the expanded marketing around these units certainly

increases the awareness and interest level about

health issues and sports in broader parts of the

citizenships. Already through the desire of

individuals in sharing and publishing their workouts

with others, these people are motivated to perform

more physical activities despite they do not have

precise measures of their efforts. Like with placebo

pills, this all can have some overall positive effect.

Another positive aspect is addressing the

opportunity of a broader exploration of futuristic UI

concepts. The four experimentation bracelets from

Fig. 1 come up with various helpful concepts for UI

handling. For instance, motion detection can assist

or even replace classical input elements like push

buttons. One of the devices, e.g., activates in this

sense its display screen, when the arm is moved up.

The other device is controllable by finger touches,

while knocking and wiping actions offer a high

dimensionality for obtaining a flat input control

hierarchy. The different output display systems

concepts will also teach, which method is acceptable

in certain environments and which is not. Variant of

the latter certainly will disappear from the market,

like the LED number in one of the bracelets that is

unreadable because the letters are too small and are

not bright enough for being readable during

daylight.

For non-technical bracelet users, the UI handling

through the website of the device vendor may appear

in the style of community pages. Also the concept of

uploading and handling all data through a central

instance on the Web follows the current data cloud

philosophy, but the approach invokes several severe

disadvantages. First of all, the UI system cannot be

used at all without Internet connection. Due to the

upload and download cycle, several instances on the

personal computer of the user has to communicate

more vulnerable to security attacks.

For the user it is also totally unclear, who all will

have in the end access to the personal data in this

system. In logic consequence to the limited motion

tracking by acceleration sensors as seen above, the

introduction of GPS traces in the bracelet will

furthermore increase privacy concerns, because then

details on location and places, where the user stays

will go to a central Web instance. This appears even

critical, because the trackers are indented for all-day

use. The Web-based UI is appears also rather poor,

if more than just vague summaries about the recent

activities shall be displayed in detail. This all stands

in full contradiction to the standalone software for

devices like T, which grant full access to all details

of individual workouts with very few selection

actions on the UI. Another negative aspect is, that

the bracelet vendors use their UI tools and the

required e-mail for uninvited information and

advertisement.

In total, the use experience of the different UI

system in the experiments here has shown, that it is

very complicate or partially impossible to access the

information details of workouts or activity traces

with the bracelet systems. All the findings here can

be summarized in the sense that the bracelet devices

in their actual construction and handling are not

professionally usable, neither in sports nor in

medical or health scenarios. For the latter - if fields

like elderly care of emergency automatisms are to be

addressed - such mal-functional systems could even

cause disastrous consequences. For ambitious and

professional sports tracking that the before

established system concepts are still serving the

requirements to a sufficient extend.

5 CONCLUSIONS

Modern electronics together with micro computer

and sensor technologies provide opportunities for

valuable handheld devices in sports and health

applications. This has been shown over many

decades also with the entry of economic commercial

devices, e.g. for measuring blood pressure or

monitoring and controlling sports activities. Such

devices can be used standalone or together with a

personal computer without Internet connection,

while producing reliable measures and traces of the

physiological activity information of interest.

For the new bracelet device class, which is also

intended and offered for the related purpose of

tracking body movement and HR, the vendors

started to enforce a totally new UI handling concept.

The user can not use the full capabilities of such

devices without Internet access, instead all data has

to be handled through Web based systems. Even

more - at least some of the devices - do not seriously

measure data, but invent data scans randomly with

the goal of exposing always nice and indicative

activity traces and functional plots in their

overloaded Web screens.

Furthermore, the user is spamed via e-mail and

while using the Web-based UI of the systems with

advertisements of alternative products. The main

benefit of the wristband systems seems to serve a

new market not in the sense of seriously providing

icSPORTS 2016 - 4th International Congress on Sport Sciences Research and Technology Support

132

any reliable measurement tools for sports and health,

but for earning quick and big profit with playful

devices. These are strategically advertised like a

fashion trend and come with general disclaimers

about their usefulness as measurement utility.

For people, who seriously want to use electronic

tools for sports and health tracking, the prior

generation of computerized handheld devices

appears at the current product state as the more

appropriate one. In contradiction to that, the new

bracelet class certainly will provoke confusion and

misguiding in sports and health use, and in health

emergency scenarios their application could even

end in disastrous situations.

REFERENCES

Alsulami, M. H., Atkins, A. S., 2016. A model of a

proactive ambient assisted living system to monitor

elderly people in the Kingdom of Saudi Arabian.

IJRET, vol. 4, no. 1 (March 2016), pp 35-41.

Arts, F. J., Kuipers H., 1994. The relation between power

output, oxygen uptake and heart rate in male athletes.

Int. J. of Sports Medicine, Vol. 15, No. 5, pp228-

231.Moore, R., Lopes, J., 1999.

Church, T., 2010. Physical Activity and Heart Disease in

Obesity, in: Physical Activity and Obesity, 2nd ed.,

Human Kinetics, Champaign (US), p282.

Colagiuri, S., et al., 2010. The cost of overweight and

obesity in Australia. Med J Aust, 192, 5 (March 2010),

pp 260-264.

Hoppeler, H., Howald, H., Conley, K., Lindstedt, S. L.,

Claassen, H., Vock, P., Weibel, E. R., 1985.

Endurance training in humans: aerobic capacity and

structure of skeletal muscle. J. of Applied Physiology,

Vol. 59, No. 2, pp 320-327.

Law M. R., Wald N. J., Meade T. W., 1991, Strategies for

prevention of osteoporosis and hip fracture. British

Medical Journal, Vol. 303, pp 453–459.

Weghorn, H., 2014. “Accuracy Evaluation of Commercial

Motion Sensing Devices in Respect to Health Activity

Monitoring Concepts,” in Proceedings of the IADIS

International conference e-health 2014, pp. 271-278,

Lisbon, Portugal (July 2014).

Weghorn, H., 2015. Application and UI Design for

Ergonomic Heart Rate Monitoring in Endurance

Sports: Realizing an Improved Tool for Health and

Sports Activities on Base of Android Smartphone

Programming and ANT+. In International Congress

icSPORTS 2013, revised selected papers, CCIS 464,

Springer Berlin, pp25-41.

Valentín, G., Howard, A. M., 2013. Dealing with

Childhood Obesity: Passive versus Active Activity

Monitoring Approaches for Engaging Individuals in

Exercise. In Proceedings BRC 2013 (Rio de Janeiro,

Brazil, February 2013).

Van Arsdale, A., 2015. Not the ideal Heart Rate tracking

tool, https://www.amazon.com/gp/review/R15S3GDU

59CQ02?ref_=glimp_1rv_cl (March 2015), last

access: 07/23/2016

Wannamethee, G, Shaper, A. G., 1992. Physical activity

and stroke in British middle aged men. British Medical

Journal, 304 (March 1992), pp597-601.

Experimental Investigation of the Usfulness of Bracelet Trackers in Sports and Health Monitoring - Critical Evaluation of a New Handheld

Activity Monitoring Device Class

133