Content uploaded by Fabio Di Francesco
Author content
All content in this area was uploaded by Fabio Di Francesco
Content may be subject to copyright.
Content uploaded by Fabio Di Francesco
Author content
All content in this area was uploaded by Fabio Di Francesco
Content may be subject to copyright.
Textile sensors to measure sweat pH and sweat-rate
during exercise
Shirley Coyle, Deirdre Morris, King-Tong Lau,
Dermot Diamond
CLARITY, NCSR, School of Chemical Sciences
Dublin City University, Dublin 9, Ireland
Nicola Taccini
Smartex s.r.l
56023 Navacchio, Italy
Daniele Costanzo, Pietro Salvo
Centro Interdipartimentale “E. Piaggio”
Università di Pisa, Italy
Fabio Di Francesco
Dipartmento di Chimica e Chimica Industriale
Università di Pisa, Italy
Maria Giovanna Trivella
Istituto di Fisiologia Clinica C.N.R.
Pisa, Italy
Jacque-André Porchet, Jean Luprano
CSEM SA
Neuchâtel, Switzerland
Abstract— Sweat analysis can provide a valuable insight into a
person’s well-being. Here we present wearable textile-based
sensors that can provide real-time information regarding sweat
activity. A pH sensitive dye incorporated into a fabric fluidic
system is used to determine sweat pH. To detect the onset of
sweat activity a sweat rate sensor is incorporated into a textile
substrate. The sensors are integrated into a waistband and
controlled by a central unit with wireless connectivity. The use of
such sensors for sweat analysis may provide valuable
physiological information for applications in sports performance
and also in healthcare.
Keywords-sweat analysis, wearable sensors, smart textiles,
personalised healthcare
I. I
NTRODUCTION
The primary purpose of the sweating mechanism is thermo-
regulation. Throughout times of physical exertion sweat rate
is increased in order to avoid a dangerous rise in body
temperature caused by the persons’ increased metabolic rate
[1]. Monitoring sweat composition in real-time can provide
useful physiological information. Sweat analysis is a valuable
diagnostic tool, being the gold standard for Cystic Fibrosis
diagnosis [2]. Physiological monitoring using sweat has the
advantage of being non-invasive and easily accessible there
are many cases where such monitoring is beneficial.
Continuous physiological monitoring of athletes during
training is a vital tool in improving performance while also
assessing the health status of the individual. Athletes must
ensure electrolyte balance and adequate re-hydration after
exercise or risk reduced performance [1, 3]. Re-hydration is a
major part of the recovery process after exercise. It is well
established that performance of exercise in a dehydrated state
is impaired, and that both high-intensity and endurance
activities are affected. There is also an increased risk of heat
illness in individuals who begin exercise in a dehydrated state
[4]. The concentration of electrolytes in sweat varies between
individuals and therefore it would be of great advantage to
develop personalized rehydration strategies depending on fluid
and electrolyte losses. Another scenario where electrolyte
balance is critical is in cases of extreme heat and/or physical
exertion where drinking too much water can lead to
hyponatremia. This is characterized by low levels of sodium
and can result in symptoms such as headache, nausea,
vomiting and muscle cramps. When there is a quick onset of
hyponatremia, for example during prolonged exercise, it can
lead to more severe complications such as seizures, coma,
brain damage and death [5]. Dehydration and electrolyte
balance is not only of vital importance to athletes but can
seriously affect vulnerable populations, e.g. in the elderly
dehydration can cause serious illness and even death
particularly in the event of a heat-wave [6]. Continuous
analysis of sweat may be used to detect changes in sweat
composition and be used as a warning indicator.
It is known that sweat pH and electrolyte concentrations
are closely related [7]. It has also been shown that sweat pH
during exercise will change with the onset of metabolic
alkalosis [8] which means that pH measurements may be used
to relate the build-up of acid in muscle cells during exercise
which leads to muscle fatigue.
While sweat may be easily obtainable, collection and
measurement techniques for analysis can be awkward -
Minor’s method involves covering the skin with starch-iodine
powder that exhibits purple dots when sweat droplets appear
[9]. “Wash-down” techniques involve exercising within a
plastic enclosure to collect sweat and then washing down the
body within the enclosure using de-ionised water at the end of
the exercise trial [10]. These methods are obviously unsuitable
for long-term measurements outside of a laboratory setting.
Parafilm patches have been used to create a capsule on the skin
surface but these may change the sweat composition by
preventing water evaporation which may block sweat gland
ducts and lead to a progressive fall in sweating rates [6].
Bioimpedance measurements have been used to estimate the
amount of water in the body. As a wearable system this is
challenging due to complex circuitry, measurement times [11]
and the effects of movement and electrode placement [12]. to
In the field, the most practical way for athletes to monitor their
sweat loss is to measure changes in body weight pre- and post-
exercise [13]. Therefore it is clear that a method is needed to
collect and analyse sweat in an unobtrusive way in normal
settings, at home, in the gym or on the track to provide real-
time analysis of sweat activity.
In addition to the analysis of sweat composition for such
applications, it is also of benefit to examine sweating patterns,
i.e. sweat rates on different regions of the body. This may be
used to identify autonomic nervous system dysfunction that
may occur due to condition such as diabetes [14]. Monitoring
sweat activity may also be useful in understanding and treating
a condition known as hyperhidrosis which causes excessive
sweating [14].
The aim of this work is to develop textile-based sensors
capable of performing chemical measurements on body fluids.
Sweat must be collected, delivered to the sensor surface and
waste products removed. This paper presents part of the
research carried out during the BIOTEX project, an EU-funded
FP6 project with a consortium of 8 partners, (see www.biotex-
eu.com). The goal has been to create a system that can be
easily integrated into fabric and for the first time perform
chemical sensing within a textile substrate.
II. M
ETHODS
A. pH fabric sensor
A textile based platform with fluid handling properties is used
to collect and analyse sweat samples. The design of the device
is shown in Figure 1(b). The system exhibits a passive
pumping mechanism, controlling fluid flow through a fabric
substrate. A fabric fluidic channel is created by screen-
printing an acrylic hydrophobic paste either side of the fabric.
At one end of this channel is a collection “window” where
sweat enters from the skin, at the opposite end of the channel a
super-absorbent (SAB) material is placed to increase the
capillary flow through the fabric and store the collected sweat.
An acquisition layer of moisture wicking fabric is stitched in
place on the skin side of the fabric patch, beneath the sensing
layer. The acquisition layer in contact with the skin helps to
increase the collection capacity of the device.
A pH sensitive dye is immobilised onto the fabric channel
using tetraoctyl ammonium bromide. Sweat pH can range
from pH 4-8, therefore the dye chosen for sweat analysis is
bromocresol purple (BCP, pKa = 6.2) which exhibits a colour
change from yellow to blue over this pH range. A paired
emitter-detector LED configuration is used to quantify the
colour change of the dye. The photo detector is a reverse
biased LED and the discharge time due to stray capacitance is
measured using a microcontroller. This technique has been
discussed previously [15]. This technique is chosen as it is a
low-power digital technique for optical detection and LEDs
are versatile components that may be easily integrated into
textiles. Two LEDs are embedded in silicone to prevent
condensation from sweat and positioned over the fabric
channel using a protective black polymethyl methacrylate
(a)
(b)
pH-sensitive
dye Hydrophobic
Coating
Optical Detection
Silicone
Gasket
Sweat SAB
Opaque cover
Silicone
Casing
Acquisition
Layer
pH-sensitive
dye Hydrophobic
Coating
Optical Detection
Silicone
Gasket
Sweat SAB
Opaque cover
Silicone
Casing
Acquisition
Layer
(a)
Figure 1. a) pH sensing patch b) fabric passive pump design
Superabsorbent
(SAB) placement
Placement for
silicone gasket
(b)
Placement area for
additional sensors
pH sensor
placement
40
mm
55 mm
20mm
Figure
2
.
a) wearable humidity sensor b) textile
arrangement to measure
low sweat rate values
(PMMA) cover held in place by a rubber gasket.
B. Sweat rate sensor
The method used to continuously measure sweat rate is to
derive its value from the gradient of humidity measured by a
pair of wearable humidity sensors at two distances (0.5 and 1.5
cm) from the skin. In cases where sweat rates are low (10-40
g/m
2
h) more accurate estimates can be obtained by increasing
the gradient. In this case, the sensor is modified by inserting a
net fabric (88% polyamide, 12% elastane) between the two
humidity sensors.
Figure 2 shows a humidity sensor and the
latter textile arrangement. The textile membrane is removed when
the measurement range has to be extended up to 1000
g/m
2
h. In
that case, a flexible textile frame is used to place the sensors at
the right positions (Figure 3). Capacitive humidity sensors are
prepared either by vacuum depositing gold electrodes onto
hydrophilic films (polyvinyl alcohol or cellulose acetate
butyrate) or by modifying a commercial device (Philips H1)
which are manufactured by a similar method.
III.
E
XPERIMENTAL SET
-
UP
The sensors are held in place using a waistband, designed by
Smartex, which ensures the fabric patch maintains good contact
with the skin during exercise (Figure 3). The waistband
is also
used to block ambient light, which may affect the operation of
the pH sensor. The waistband is made from a moisture
wicking, stretch fabric for comfort with an integrated pocket to
hold the pH sensing patch in place. The use of an elastic fabric
laminated with a hydrophobic membrane prevents the
absorption of sweat from the waistband and the reduction of the
sweat flow through the pump. Hooks and loops fasteners are
used to keep the passive pump and the electronics in place.
This reduces motion effects during exercise.
A second pocket houses the sweat rate sensor. A neoprene
frame makes sure that the right distance between sensors and
skin is maintained, while holes in the pocket allow the water
vapour to diffuse away without obstacles. A reference patch
for pH measurements is also included in the waistband. This is
used to make pH measurements via a Skincheck
TM
pH meter,
which is pressed against the patch at 5 minute intervals.
Reference measurements are taken once the subject is observed
to begin sweating. The waistband is positioned on the subjects
back so that both the pH sensor patch and the reference patch
are to the right of the spine. The sweat rate sensor is positioned
to the left of the spine. This is done to avoid collecting sweat
which runs from the top to the bottom of the back along this
hollow from other parts of the back.
The LED optical detection configuration is controlled by a
central unit developed by CSEM (www.csem.ch). The detected
light intensity relating to pH measurements is transferred
wirelessly to a nearby laptop using a Bluetooth communication
module within the central unit. The control unit also interfaces
with the sweat rate sensor by measuring the capacitance of each
humidity sensor. Both pH and sweat rate measurements are
sampled at a rate of 0.2 Hz. The exercise set-up is shown in
Figure 4. The protocol involved indoor cycling in a room at
20°C with relative humidity of 50%. Subjects were asked to
cycle for up to 60 minutes at a self-selected pace.
IV. R
ESULTS
Results of exercise trials on two healthy male subjects are
shown in Figure 5. Sweat rate is represented by the difference
in capacitance between the two humidity sensors. Sweat rate
data have been normalised and smoothed using a moving
average of 12 points. Figure 6 shows how the sensor signal
follows the variations of the sweat rate measured by a
commercial reference meter (Vapometer, Delfintech) quite
well. The Vapometer is currently used in many fields such as
Sweat Rate sensor
housing
(a)
(b) (c)
pH sensor pocket
Reference patch
window
Figure 3. a) the waistband b-c) details of the housings of the pH sensing
patch and sweat rate sensor
Sweat rate
sensor Control
electronics
Figure 4. Experimental set-up, sweat rates sensor is positioned on the
left, the pH textile patch on the right behind the control unit
academic skin research and testing laboratories, personal care
and chemical industries for the measurement of evaporation
rates.
The temporal characteristic of sweating pattern is useful for
interpreting the response of the pH sensor. A priming time is
involved before the pH measurements are valid. This involves
the time it takes for the subject to start sweating and for the
sweat to enter the fluidic channel. This varies between
individuals and depends on fitness levels and environmental
conditions. The priming time is also indicated by the
reference measurements as these measurements are taken once
there is adequate sweat on the reference patch to wet the
reference electrode. From the results in Figure 5 this priming
time corresponds to the time taken to reach a constant sweat
rate as measured by the sweat rate sensor. Once the sweat
enters the channel it takes up to 5 minutes before a complete
colour change in the pH sensitive dye is achieved. Sweat pH
is observed to increase as the exercise trial progresses and as
the sweat rate increases. It can be seen that for both trials
there is close agreement between the results of the textile
based pH sensor and those obtained using the reference
method. These data have demonstrated that the performance of
the textile sensor is comparable to the commercial system.
Differences may be attributed to the fact that the pH sensor
and reference patch are sampling sweat from separate regions
of the back [7]. The increase in pH during exercise can be
explained by considering the anatomy of sweat glands and
their operation. Sweat glands have two main components, a
tubular coiled area where the sweat is produced and the duct
through which it reaches the skin surface. The primary sweat,
formed in the secretory coil is near iso-osmotic, with a pH
close to 7, while that excreted onto the skin is hypo-osmotic,
making it acidic. It is thought that this is most likely due to
ductal transport processes, which are dependent on the
physiological condition [8]. The sweat becomes acidic due to
re-absorption processes which occur in the duct. However, as
the subject continues to exercise, the rate at which sweat is
excreted increases, in order to regulate body temperature. This
reduces the amount of time available for re-adsorption
processes and sweat pH increases.
V. D
ISCUSSION
Sweat is a clear hypotonic, odourless fluid containing
sodium, chloride, potassium, urea, lactate, bicarbonate,
calcium, ammonia, organic and non-organic compounds [16].
Clearly there are many different constituents in sweat that are
inter-related. pH is a useful parameter to measure as every
function of the body is dependent on our bodies maintaining a
precisely balanced pH in order to function correctly. The
enzyme system, as well as the electrical functions of our body,
is dependent on electrolytes. Rehydration after exercise
requires not only replacement of volume losses, but also
replacement of the electrolytes, primarily sodium, lost in the
sweat. Sweat composition not only varies among individuals
but also varies with time during exercise, and is further
influenced by the state of acclimatization [4]. The onset of
significant dehydration is preventable, or at least modifiable,
when hydration protocols are followed to assure all athletes
the most productive and the safest athletic experience [17].
Therefore individual monitoring would enable personalized
rehydration drinks to be prepared for each athlete to meet their
needs. The design of a wearable sweat analysis system has
been demonstrated as a step towards such strategies. Further
trials with a wider variety of subjects in different training
scenarios are needed to further validate the system.
The fabric pH sensor has been demonstrated during indoor
cycling trials. To function it needs an adequate quantity of
sweat entering the patch, which often takes up to 20 minutes
for the person to start to sweat. Simultaneous measurement of
sweat rate is useful to know when the pH readings are valid.
0
1
2
3
4
5
6
7
0 10 20 30 40 50 60
Time (min)
pH
pH sensor
reference
0
0.2
0.4
0.6
0.8
1
0 10 20 30 40 50 60
Time(min)
Sweat rate (a.u.)
Priming
time
pH measurement
0
1
2
3
4
5
6
0 10 20 30 40 50
Time (min)
pH
pH sensor
reference
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 10 20 30 40 50
Time(min)
Sweat rate (a.u.)
Priming
time
pH measurement
Figure 5.
Results of exercise trials on two healthy volunteers: sweat rate and pH values measured during a 60 minutes effort at self
-
selected pace
The priming time of the sensor will vary depending on the
individual and their sweating patterns. Future work will aim
to reduce the priming time by reducing the sensor surface area
where a lower quantity of sweat would be required for the
reaction to occur. The dimensions of the current design were
chosen to accommodate the placement of additional sensors on
the channel so that more detailed analysis of sweat
composition could be performed. The work presented here is
just part of work involved in the BIOTEX project that tackles
the issues involved in real-time sweat analysis. To get a
complete picture the overall BIOTEX system includes sodium,
chloride and sweat conductivity sensors along with
measurement of other physiological parameters such as
breathing, heart rate and blood oxygen saturation. The sweat
sensors may provide complimentary signals to ECG and
respiration data which have already been demonstrated using
textile sensors [18, 19] to form a complete wearable healthcare
monitoring garment. By developing textile sensors, the
sensing elements can be easily integrated and distributed
throughout a garment. This means that to study sweating
activity patterns the sweat rate sensors could be placed at
different regions of the body which may assist in diagnosis
and treatment of illnesses affecting the autonomic nervous
system. The great advantage of textile sensors is that they
enable long term monitoring of the wearer at home, in normal
situations.
VI. C
ONCLUSIONS
Textile based sensors to measure sweat pH and sweat rate
during exercise have been demonstrated. Further work is
needed to relate the sweat rate measurements with flow rates.
This could provide valuable information to athletes regarding
their rehydration needs and also in clinical applications for
treatment and diagnosis of conditions such as hyperhydrosis.
The temporal response of sweating patterns provided by the
sweat rate sensor has been useful in interpreting the response
of the fabric pH sensor which needs adequate sweat rates to
function effectively. Further work is needed to investigate
other constituents of sweat such as electrolyte concentrations
and also investigation of different training regimes in different
environmental conditions. Analysis of these different
parameters should help to understand the sweating mechanism
and open a new area of research as a non-invasive approach to
physiological measurements.
This area of research is unfortunately lacking due to the
difficulty in sampling sweat. It is hoped that this system will
open a new avenue of research allowing physiologists to
explore the benefits of non-invasive sweat studies. Textile
sensors in contact with the skin offer much promise to provide
novel physiological information. The textile based fluid
handling system that has been developed provides a platform
for implementing wearable biological and chemical sensors.
Garments integrating such textile sensors allow innocuous
monitoring of the health and well-being of their wearers. This
has many applications in sports performance and also in
healthcare.
A
CKNOWLEDGMENT
We gratefully acknowledge the financial support of the
European Union (Biotex FP6-2004-IST-NMP-2) and Science
Foundation Ireland (07/CE/I1147).
R
EFERENCES
[1] S. J. Montain, "Hydration recommendations for sport 2008," Current
sports medicine reports, vol. 7, pp. 187-192, 2008.
[2] J. Massie, K. Gaskin, P. Van Asperen, and B. Wilcken, "Sweat testing
following newborn screening for cystic fibrosis.," Pediatric
Pulmonology, vol. 29, pp. 452-456, 2000.
[3] Australian Institute of Sport, "Fluid - who needs it?," 2004.
[4] R. J. Maughan, J. B. Leiper, and S. M. Shirreffs, "Rehydration and
Recovery after Exercise," Sports Science Exchange, vol. 9, pp. 1-5
Supplement 62, 1996.
[5] H. J. Adrogue and N. E. Madias, "Hyponatremia," Primary Care, vol.
342, pp. 1581-1589, 2000.
Figure 6.
Response of textile sweat rate sensor compared to commercial Vapometer
[6] S. M. Shirreffs and R. J. Maughan, "Whole body sweat collection in
humans: an improved method with preliminary data on electrolyte
content.," Journal of Applied Physiology, vol. 82, pp. 336-341,
1997.
[7] M. Patterson, S. Galloway, and M. A. Nimmo, "Variations in regional
sweat composition in normal human males," Experimental
Physiology, vol. 85, pp. 869-876, 2000.
[8] M. Patterson, S. Galloway, and M. A. N. Immo, "Effect of induced
metabolic alkalosis on sweat composition in men," Acta Physolol
Scand, vol. 174, pp. 41-46, 2002.
[9] K. T. Sato, A. Richardson, D. E. Timm, and K. Sato, "One-step iodine
starch method for direct visualization of sweating," American
Journal of Medical Science, vol. 6, pp. 528-531, 1988.
[10] P. W. Lemon and K. E. Yarasheski, "Feasibility of sweat collection by
whole body washdown in moderate to high humidity
environments.," International Journal of Sports Medicine, vol. 6,
pp. 41-43, 1985.
[11] T. Vuorela, K. Kuttonen, J. Rantanen, T. Jaervinen, and J. Vanhala,
"Bioimpedance Measurement System for Smart Clothing,"
presented at Proceedings of the Seventh IEEE International
Symposium on Wearable Computers, 2003.
[12] G. Medrano, L. Beckmann, N. Zimmermann, T. Grundmann6, T.
Gries, and S. Leonhardt, "Bioimpedance Spectroscopy with textile
Electrodes for a continuous Monitoring Application " presented at
4th International Workshop on Wearable and Implantable Body
Sensor Networks (BSN 2007) March 26 – 28, 2007 RWTH
Aachen University, Germany 2007.
[13] R. J. Maughan and S. M. Shirreffs, "Development of individual
hydration strategies for athletes," International Journal of Sport
Nutrition and Exercise Metabolism, vol. 18, pp. 457-72, 2008.
[14] M. Asahina, Y. Yamanaka, Y. Akaogi, S. Kuwabara, Y. Koyama, and
T. Hattori, "Measurements of sweat response and skin vasomotor
reflex for assessment of autonomic dysfunction in patients with
diabetes.," Journal of Diabetes Complications, vol. 22, pp. 278-83,
2008.
[15] K. T. Lau, W. S. Yerzunis, R. L. Shepherd, and D. Diamond,
"Quantitative colorimetric analysis of dye mixtures using an
optical photometer based on LED array," Sensors and Actuators B,
vol. 114, pp. 819-825, 2006.
[16] A. G. R. Whitehouse, "The Dissolved Constituents of Human Sweat,"
Proceedings of the Royal Society of London, vol. 803, pp. 139-154,
1935.
[17] D. Casa, L. E. Armstrong, S. Hillman, S. J. Montain, R. Reiff, B. S. E.
Rich, W. O. Roberts, and J. Stone, "National Athletic Trainers’
Association Position Statement:Fluid Replacement for Athletes,"
Journal of Athletic Training, vol. 35, pp. 212-224, 2000.
[18] R. Paradiso, G. Loriga, and N. Taccini, "A Wearable Health Care
System Based on Knitted Integrated Sensors," IEEE Transactions
on Information Technology in Biomedicine, vol. 9, pp. 337-344,
2005.
[19] M. Engin, A. Demirel, E. Z. Engin, and M. Fedakar, "Recent
developments and trends in biomedical sensors," Measurement,
vol. 37, pp. 173-188, 2005.
