0% found this document useful (0 votes)
14 views28 pages

Review of Stress Detection Methods Using Wearable Sensors

Uploaded by

Ondra Fater
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
14 views28 pages

Review of Stress Detection Methods Using Wearable Sensors

Uploaded by

Ondra Fater
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 28

Received 14 December 2023, accepted 21 February 2024, date of publication 4 March 2024, date of current version 15 March 2024.

Digital Object Identifier 10.1109/ACCESS.2024.3373010

Review of Stress Detection Methods Using


Wearable Sensors
GEORGIOS TASKASAPLIDIS 1 , DIMITRIS A. FOTIADIS 1, AND
PANAGIOTIS D. BAMIDIS 2 , (Member, IEEE)
1 Informatics Department, School of Sciences, University of Western Macedonia, 52100 Kastoria, Greece
2 School of Medicine, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
Corresponding author: Georgios Taskasaplidis ([email protected])
This work was supported in part by the University of Western Macedonia, Greece.

ABSTRACT Stress is a significant factor that affects well-being and health. Factors that trigger stress
include work, social interactions, and economic and environmental factors. Stress may cause lower
labor productivity, physical and mental health problems, and malfunctions in all social aspects of life.
Psychosomatic health can be improved if proper stress detection mechanisms are present in daily life and
stress reduction methods can occur. Wearable sensors are currently used in many commercial and scientific
applications in a non-invasive or unobtrusive manner. These devices are used in daily routines. In this paper,
a comprehensive review of the latest literature and developments in stress detection methods is presented
through extensive and holistic research on stress response, both at the level of the autonomic nervous system
(ANS) and hypothalamic-pituitary-adrenal axis (HPA). This review focuses on the exploitation of various
methods, technologies, and data analysis systems to understand stress in a multifaceted and comprehensive
manner. Various stress-related factors are presented along with biological signal measurements, and physical
secretions or biomarkers are primarily used for stress detection. Furthermore, the manner in which body
movement and posture measurements may be related to stress was investigated, together with speech and
hand tremors. Various stress-detection technologies have been analyzed, and existing data analysis methods
that can be applied to stress-detection systems have been highlighted. This review serves as a reference and
guideline for exploring this area of interest, identifying research opportunities, and offering ideas, options,
and suggestions for optimized solutions regarding future applications and research.

INDEX TERMS Stress analysis, stress detection, stress response, wearable sensors.

I. INTRODUCTION The World Health Organization estimates that every dollar


Stress in modern days is considered not only an important spent on fighting depression and anxiety is repaid four
factor of well-being but also a factor of basic health. Stress times [3].
has been described as a ‘‘hidden epidemic’’ of modern times American psychologist Walter Cannon (October 19, 1871
at least since 2002 [1]. – October 1, 1945) defined stress as any deviation from
The economic impact of stress is considered to be homeostasis [4]. Homeostasis, the steady physiological state
huge, especially when discussing work stress, where it is of the living human body, was also proposed by him.
estimated that approximately $300 billion is spent annually Critical parameters, such as temperature, heartbeat, and blood
in the US [2]. pressure, can be measured within a certain limit using
biological sensors.
Many professional, social, and personal problems as
well as difficult situations are caused or aggravated by
The associate editor coordinating the review of this manuscript and stress. Labor productivity decreases significantly under
approving it for publication was Roberto C. Ambrosio . conditions of stress [5], depression, and some psychogenic
2024 The Authors. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
VOLUME 12, 2024 For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/ 38219
G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

and pathological diseases as well as in the sexual ability of and digest’’ conditions, thus relaxing the body, and at the
both men and women [6]. HPA axis level, the hypothalamus inhibits stress hormone
To solve this problem, it is important to detect their production.
presence first. Thus, the causes of the problem (stress), Historically, the polygraph (known as the lie detector
place, and time can be identified, treated, or mitigated. test) measures and monitors the breathing rate, perspiration,
However, stress, in general, is a natural physiological skin conductivity, blood pressure, and pulse rate [8]. Poly-
defensive-compensatory mechanism in the body that creates graphs can be considered an ancestor of stress-measurement
problems only when it is uncontrollable, beyond certain systems. The classical, reliable, and accurate method for
limits, and unmanageable. It should also be mentioned that measuring and quantifying psychological stress was devel-
sometimes controlled stress is deliberately caused (e.g., oped in 1983 by S. COHEN et al. (Journal of Health and
intensive intermittent exercise, cold, heat) by the person, Social Behavior, 1983) and is a questionnaire known as the
so that he later receives the beneficial, soothing, and healing Perceived Stress Scale (PSS) [9]. There were three versions
interventions of the brain to the body (hormone secretions, of the questionnaire. The original questionnaire used was the
endorphins). PSS-14. Subsequently, shorter versions, such as the PSS-4
and PSS-10, were introduced [10]. Technology has advanced,
and stress can be measured (‘‘automatically’’), even with
wearable devices.
The modern trend of using mobile devices, such as laptops,
tablets, or smartphones, and the continuous development
of technology in this field, with increasingly powerful and
smaller processors and chips that consume less energy but
also optimize batteries (in terms of capacity, charging speed,
and charge cycle), are beneficial for the development of
wearable stress detection devices.
Miniaturized semiconductor devices appear more and
more into commercial electronic devices, aided by new
microfabrication technologies [11]. These technologies allow
the development of small-sized wearable devices that can
assist in stress detection [12], [13], [14].
Wearable sensors are often used in simple and practical
applications. These sensors are friendly to the person being
investigated for stress, have low complexity, and do not
require cumbersome medical protocols. They can be easily
and safely applied to almost any person or environment,
including daily life, hobbies, or work.
Wearable devices that can be worn almost everywhere are
an attractive way to monitor stress. Watches, wristbands,
rings, bracelets, necklaces, and eyeglasses can be utilized
FIGURE 1. Stress response: Role of the hypothalamus (source: to sense stress-associated biosignals. Furthermore, straps
authors).
(chest straps), patches, lenses, and other devices that may
be less friendly but still practical can be used for this
Two main mechanisms respond to stress stimuli and are purpose.
coordinated by the hypothalamus [7]. The hypothalamus is Smartwatches are wearable devices that have evolved
a part of the brain located below the thalamus. It is also significantly in the last decade. Commercially available
associated with the nervous and endocrine systems (Fig. 1). smartwatches not only show time, month, and year but also
In response to stress, it can trigger (a) the sympathetic have many other modern features, such as Internet connec-
nervous system (SNS), which is a part of the autonomic tivity, various applications support, and integrated sensors
nervous system (ANS), and (b) hormone production. SNS for wellness. These sensors can be used to detect human
can control the so-called ‘‘fight-or-flight’’ response and, for biomarkers that can measure the heart rate, temperature,
instance, can increase heart rate, blood pressure, and eye pupil oxygen saturation in the blood, and blood pressure. These
enlargement. Additionally, the hypothalamus can provoke the measurements are sometimes accurate and certified by the US
production of hormones through the hypothalamus-pituitary- Food and Drug Administration (FDA).
adrenal (HPA) axis to counter stress. In contrast, under A thorough investigation of the latest (by priority)
stress recession conditions, the hypothalamus can trigger developments and literature concerning stress detection,
the parasympathetic nervous system (PNS), which is a part response, and measurement was conducted using a holistic
of the ANS. The PNS predominates in the so-called ‘‘rest approach. Common and uncommon biomarkers, sensors,

38220 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

devices, methods, and technologies have been explored. of several measurements may detect and expose stressful
Research has also been conducted at another (medical) situations.
level, for example, in human secretions (related to the stress
substances of these secretions), hand tremors, body posture, 1) HEART RATE (HR)
eye dynamics, and sensors that respond to and measure all of
HR increases when the body requires either more oxygen
these elements. This review is a synthesis-correlating product
or the elimination of carbon dioxide. Acute stress responses
of such information. Furthermore, the prevailing methods
increase the HR.
of data analysis and classification as well as their transfer
Heart rate variability (HRV) refers to slight time fluctua-
and transmission methods are highlighted. Several sources
tions between heartbeats and is generally not considered an
have been used to prove various cases and points. As for
arrhythmia [21]. The normal heart beating is called ‘‘sinus
essential concepts and historical stages in the study of stress,
rhythm’’, and if the variability between heartbeats is greater
older articles, books with significant impacts, and sources of
than 0.12 s, then this event is called ‘‘sinus arrhythmia’’.
sources were also used. Therefore, a broader, more complete,
Breathing usually induces sinus arrhythmia; thus, HRV can
and up-to-date view with various options, tools, and methods
be used to measure breathing rate [22]. The term R-R
was provided. This review intends to identify innovative,
interval or inter-beat interval (IBI) is the time (measured in
unexplored, and underexplored areas in this field that can
milliseconds) between heartbeats.
optimize stress detection and measurement and introduce
The sympathetic and parasympathetic systems are parts
new ideas and research horizons. Suggestions, directions, and
of the autonomic nervous system ANS [23]. The ANS is a
estimations are also provided.
subconscious mechanism of the brain and the nervous system
The remainder of this paper is organized as follows. In the
that controls the heart. The sympathetic nervous system
next section, various stress-related factors that could be
induces a faster heartbeat and a higher blood pressure. This,
used for stress detection are discussed. Section III presents
in turn, provides more oxygen to the body to cope with
a detailed description of commercially and scientifically
stressors (e.g., fear, danger, and rush). Simultaneously, the
available stress detection technologies, and various body
parasympathetic system attempts to restore all of the above
measurement points and methodologies are described. The
to normal conditions (reducing blood pressure and heart rate)
manner in which these measurements were collected from
after the stressor or initial stimulation has gone or been
a networking perspective is presented in Section IV. Var-
reduced.
ious data analysis methods are described in Section V.
Electrical changes (voltage) are observed in the heart
A discussion and overall conclusions are provided in
muscle, and a graphic representation of these electric signals
Sections VI and VII, together with a general proposal for the
represents an electrocardiogram (ECG) [24]. ECG, HR, and
most preferable solution.
HRV were easily measured. ECG is explained in more detail
as an electrical technology for stress detection in Section III
of this paper.
II. STRESS RELATED FACTORS The heart produces seismic and ballistic effects. Wearable
A. BIOLOGICAL SIGNAL MEASUREMENTS wrist devices with accelerometers (e.g., triaxial microelec-
tromechanical (MEMS) accelerometers) and electromechan-
Detecting stress in real time and analyzing biosignals from
ical film (EMFi) sensors can detect these effects [25].
wearable devices in everyday life – routine–are challenging
Seismocardiography (SCG) and ballistocardiography (BCG)
tasks. This task becomes easier in controlled environments
[26] are methods reported in the literature that can measure
(for example, in medical laboratories), especially when blood
HR and breathing rate [27] using ballistic forces and
analysis is possible together with other specialized laboratory
vibrations of the heart.
methods. On the other hand side, in the last years, out-of-
Finally, heart rate can be measured optically using
lab detection of stress is also of equal importance and the
photoplethysmography (PPG). This method measures blood
technological advances of wearables assisting in detecting
volume pulse (BVP). Recently, commercially available
stress are enormous [15].
wearable devices have been offering reliable (with FDA
The autonomic nervous system (ANS) responds to
clearance) and continuous measurements of wrist heart rate
stress [16], [17]. It has also been scientifically proven that
using the PPG method.
the ANS regulates the functions of the cardiovascular system,
skin, respiration, eye (pupil size, eye blinking, and eye
movement), and many other organs. 2) TEMPERATURE
A variety of biosignals from the human body, which are Temperature (usually in association with others) is a phys-
controlled by the ANS and therefore affected by stressful iological measurement used for stress detection [29], [30],
conditions, can be obtained through wearable devices [18], [31], and skin temperature may reveal the intensity of acute
[19], [20]. The most commonly used biological signals stress [32]. Skin temperature is usually measured rather than
originate from the heart and skin, and are probably the inner body temperature using wearable sensors for usability
most important. Individual measurements or a combination reasons. Most measurements are taken on the wrist [33], [34]

VOLUME 12, 2024 38221


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

FIGURE 3. Blood oxygen saturation, heart rate, and blood pressure were
FIGURE 2. Various wearable biosignal devices and utilities can assist with measured using a medical wristwatch via photoplethysmography and
stress detection and analysis [28]. artificial intelligence (AI) [38].

(wearable devices such as watches or watch-style devices), 4) BLOOD PRESSURE (BP)


fingers, or the main core of the body skin ( electronic Blood pressure (BP) measurements are related to the heart
patches [30], specialized belts with integrated sensors, and arteries. As blood flows through the arteries, pressure
straps, etc.). Some wearable thermometers are equipped is produced at their walls. The systolic and diastolic
with thermistors, others with embedded liquid crystal strips, pressures were measured. BP reactivity, along with other
and others use infrared (IR) emission technology [35]. pathophysiological reactions, is repeatedly associated with
High-precision measurements can be recorded using all the stress [42], [43]. However, many other physical activities can
employed technologies. increase BP, indicating that it is not a simple task to associate
Skin temperature measurement via the wrist is mostly BP with stress individually.
performed using a contact thermistor or a contactless IR ther- Some wearable wrist-based devices measure BP, and only
mometer. Normal skin temperature is usually approximately a few are clinically accurate and approved by the FDA [44],
33 ◦ C to 35 ◦ C, which is lower than the usual normal body the majority of which are not sufficiently accurate [45].
temperature (33.6 ◦ C to 37 ◦ C). Accurate BP measurement is related to stress and critical for
Changes in body temperature may indicate disease, the prognosis of hypertension. Hypertension is considered the
abnormalities, or possibly stressful conditions. most common risk factor for cardiovascular diseases [46].
Ballistocardiography and the time taken for the systolic
3) BLOOD OXYGEN SATURATION (SPO2) pressure wave to travel from the aortic valve to a peripheral
A large percentage of oxygen is transported through the site, called the pulse transit time (PTT), offer an opportunity
red blood cells (hemoglobin). Hb oxygen saturation can be to estimate cuffless blood pressure monitoring [47] with
measured by several methods. Usually, this measurement is the synergy of photoplethysmography (PPG) and/or electro-
useful for detecting hypoxia (poor oxygenation of the body, cardiography (ECG) [45]. Cuffless-estimated BP based on
particularly the brain) [36]. It is a measurement that medical bioimpedance sensors is also expected to be commercially
personnel and lay people use for diagnosis or preventive available [48].
treatment. Recently, coronavirus disease, also known as
Covid-19, sometimes has the effect of ‘‘silent’’ hypoxia,
and this measurement is now widely known in the medical
field. Wristwatches can continuously and smartphones can
instantly measure blood oxygen saturation via red and
infrared reflection-led light [37]. In the case of stress, some
possible variations in this measurement were investigated and
their correlation with it (stress).
Stress and anxiety alter the respiratory rate, thereby
altering oxygen saturation in the blood [39]; hence, there is
a correlation between the breathing rate and blood oxygen
FIGURE 4. Real-time blood pressure measurement using a commercially
saturation. available wristwatch [49].
There are three methods for measuring blood oxygen
saturation: wearable fingertip pulse oximetry, transmission
photoplethysmography (T-PPG), and reflectance photo- 5) BRAIN ACTIVITY
plethysmography (R-PPG) [40]. Only a few wearable devices The brain regulates the autonomic nervous system (ANS)
have clinical grade accuracy of SpO2 [41] with FDA or other through a region of the brain called the hypothalamus.
medical organization approval. The ANS, specifically the sympathetic and parasympathetic

38222 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

systems, responds to emotions, cognition, and mental stress. gated. Usually, a combination of measurements, such as
The hypothalamus controls many vital functions such as heart EDA (GSR), EMG, electrocardiography (ECG), EMG [61],
rate, body temperature, thirst, and sleep and responds to or other combinations, are used to determine, evaluate,
an ANS arousal alert with hypothalamic hormones to bring or investigate stress responses.
about body balance (homeostasis) [50].
Neurons in the cerebral cortex produce electricity (current 7) RESPIRATION RATE AND VOLUME
ions) or voltage fluctuations. Wearable devices are used in Respiratory breathing involves the inhalation and exhalation
scientific studies and fitted to the head (scalp) to measure of air to and from the lungs. Respiration rate is the number
and record this activity [51]. However, these devices are of breaths per unit of time, typically per minute. In stressful
difficult to wear and can be worn widely in any environment situations, this rate changes [62].
in a manner similar to devices on the wrist, chest, arm, and Breathing rate can be associated with heart rate. Specif-
skin. Nevertheless, they are becoming increasingly attractive ically, it is associated with heart rate variability [21].
(lighter, with less volume, and easier to apply), particularly at Breathing rate can also be measured using electrodes [63]
home and in the office.
(through electrical fluctuations and impedance [64]), radio-
Near-infrared (NIR) light emitted (from a device) to
frequency sensors [65], piezoelectric sensors [66], [67], and
the outer cortex of the brain is absorbed, resulting in
strain sensors [68]. Currently, several commercially available
changes in quantity upon receiving a stimulus. Thus, NIR
wearable devices are in use.
methods or devices on the scalp can be used to study these
Measuring respiratory rate using wearable devices is
functional signals [52]. Wearable devices can be used for NIR
relatively easy, and there is a plethora of wearable devices
spectroscopy [53]. Evaluation, detection, and measurement of
that perform such tasks. Measuring breath volume is more
stress via brain activity [54] as the sole factor – parameter [55] difficult, but steps are being taken in this direction [65],
or in combination with other activities such as biosignals [56] [68]. Usually, under stress conditions, a fast respiration rate
of the human body (e.g., skin and heart). induces shallow breathing [69].
Breathing can have the opposite effect [70]. Control of
6) DERMAL AND MUSCLE ACTIVITY the respiration rate and depth (volume) can reduce blood
pressure [71] and heart rate variability [72], which can
The skin is the largest in the human body. The skin covers
lead to relaxation and stress reduction by enabling the
the outer part of the body. Among many other things, the skin
regulates body temperature with a cooling system commonly parasympathetic system (vagus nerve) [73].
known as sweating. Sweating is controlled and triggered by
the sympathetic nervous system and alters the conductivity 8) EYE DYNAMICS
and humidity of the skin surface [57]. Electrodermal activity The pupillary dilator muscle of the iris is controlled by sym-
(EDA) is the ability of the skin to vary its electrical pathetic and parasympathetic systems. The pupil responds
conductance and electrical characteristics. EDA is frequently (a) to light (pupil light response (PLR)), (b) to the near
used in psychophysiological evaluations [58] and stress. The fixation-near reflex (pupil near response (PNR)), and there
skin conductance (SC) and skin potential (SP) response peak is another response (c) to the pupil psychosensory response
time difference between these signals (SC and SP) during (PPR) related to cognitive activity [74]. The last response of
relaxation and stress (or an unexpected cause) [59]. The skin the pupil is related to mind effort or mind processing load
is directly affected by the human sympathetic system, and no after something has captured attention or strong emotion and
parasympathetic action directly affects the skin; thus, it is an causes pupil dilation through the sympathetic system.
ideal part of the human body to observe the action, stimulus, The sympathetic system responds to stress and induces
and effect of the sympathetic system. pupil dilation, whereas the parasympathetic system stimula-
The galvanic skin response (GSR), which is part of the tion causes pupil constriction [75]. Therefore, pupillometry
EDA, detects changes and fluctuations in the electrical is another possible method to detect and evaluate stress.
response of the skin sweat glands. GSR are formed by Eye blinking [54] or eye closure is another additional
the synthesis of phasic and tonic components [60]. The measurement that uses an eye tracker (more often, specialized
differences between tonic and phasic are in terms of time and glasses are used) that can be evaluated or co-evaluated for
abrupt variations. The phasic part exhibited sharp outbursts stress detection [76]. In addition, eye gaze [77], eye move-
and short responses (a few seconds), whereas the tonic part ment (rapid movements), and eye activity, in general [20],
was typically slower (seconds to minutes) and smoother. [78].
Muscle cells produce electric potentials when they are However, eye measurements require specialized and
neurologically or electrically activated. Electromyography expensive devices. These measurements are ideal for individ-
(EMG) can measure this activity using EMG electrodes on uals who can wear helmets at work, or during other activities.
a muscle (e.g., trapezius muscle). The relationship between For example, eye dynamics measurements can be performed
EMG activity and increased activity after a stimulus or on pilots (airplanes, racing cars) [80] and truck drivers [20],
stressor in some muscles has been evaluated and investi- so they will not fall asleep, their attention is low, they have a

VOLUME 12, 2024 38223


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

FIGURE 5. Eye tracker glasses for stress detection with cameras for eye
pupil diameter and blinking frequency observation, and other FIGURE 6. Stress response: The hypothalamus releases corticotropin
sensors [79]. hormone (CRH) and triggers the release of adrenocorticotropic hormone
(ACTH) from the anterior pituitary into the circulation. Finally, the adrenal
cortex releases stress hormones (such as cortisol) [94].

high emotional workload, etc. Novel devices such as glasses


with eye trackers, are now easier to implement and are used Human physiological fluids that can be measured using
in a number of applications. noninvasive Electrochemical Biosensors include saliva, sweat
(eccrine sweat and apocrine sweat), tears, interstitial fluid
B. PHYSICAL SECRETIONS - BIOMARKERS (ISF), and urine [86], [88], [89], [95], [96], [97].
In the previous chapter, the role of the ANS, which
consists of the sympathetic and parasympathetic systems in 1) SWEAT
response to stress, and the role of the hypothalamus as a Sweat is a salty fluid consisting mainly of water secreted from
regulatory (or as a control center) system of the ANS were the skin-sweating glands of mammals. Eccrine and apocrine
discussed. The hypothalamus is also ‘‘communicating’’ with sweat glands are also observed. Eccrine sweat glands are
the pituitary gland and adrenal cortex via hormones (Fig. 6). present throughout the body and play a thermoregulatory role
The organs and interactions among these three components (through the evaporation effect of sweat). Apocrine sweat
are known as the hypothalamic-pituitary-adrenal (HPA) axis. mainly exists in the hair follicles in the groin and axillary
Therefore, the HPA axis is a major neuroendocrine system areas. Apocrine sweat is (more) osmotic, and apocrine sweat
that interacts with the nervous and endocrine systems, glands produce thicker sweat with a lower percentage of
controls stress responses [81], and maintains physiological water than eccrine sweat [98], [99].
homeostasis [82]. The following biomarkers [100], [101] can be identified:
Electrochemical sensors detect the chemical, physical, lactate [102], ammonia, calcium, glucose [93], [103], sodium,
or biological parameters [83], [84] of a part of the human chlorine, potassium, ethanol, urea, cortisol [104], [105],
body (including body fluids, epidermis, specific body areas, [106], and various cytokines and neuropeptides.
and places). They can detect hormones, enzymes, and The proper method of sweat sample collection, with
proteins produced by the HPA axis. Recently, such sensors an emphasis on preventing the sampling of mixed fluids
have been developed in a wearable form [85], [86], [87], and elements from other sources on the skin surface, can
[88], [89]. Wearable devices are gaining increasing attention accurately reflect glucose [20] and cortisol [107] levels in
for monitoring physicochemical parameters (e.g., HR, EDA, correlation with saliva and blood. Cortisol and glucose levels
motion, respiration rate, and skin temperature), and a in response to stress.
combination of biomarker monitoring is required at the next Several sensors that can measure all aforementioned
level of a more complete and sophisticated overview of health elements are available [86], [101], [108], [109], [110].
and body functioning [88]. Many of these biomarkers are Many of these are wearable [111], [112], [113], [114],
involved in stress detection, monitoring, and evaluation. [115], whereas some are wireless [89], [106]. All these
Measurements are mainly performed on saliva [87], [90], sensors have been developed and are continuously being
[91], and eccrine sweat [91], [92], [93], and less on other developed, are researched and commercially available mainly
secretions. Both (Saliva and sweat) are attractive alternatives for athletic and medical interests, and focus on sweat
for blood analyses. Some elements found in saliva and eccrine measurements [116], [117].
sweat were correlated with those found in blood. Skin is the largest human organ and has the largest surface
These biomarkers are cortisol (hormone), glucose, pro- area throughout the body. It provides continuous real-time
lactin (protein), and alpha-amylase (enzyme), which are detection, measurement, and quantification of biomarkers
usually detected, compared, analyzed, and related to stress. using wearable sensors, making it much more practical,
Stress-related biomarkers have been described in detail in this convenient, and less annoying than other bodily fluids (e.g.,
section. saliva, urine, and tears).

38224 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

2) TEAR
Tears are salty liquid fluid that is secreted from the tear glands
(lacrimal glands) and found in the eyes. They contain water,
electrolytes (which is why they are saline), proteins, lipids,
and other elements. The tear film consists of three layers.
From the lipid, aqueous, and mucous layers [118]. The main
role of tears is to lubricate the eye, remove irritants (reflex
tears), and provide immune support [119]. Pain and emotions
can ‘‘bring’’ tears in the eyes [120].
Tear fluid sampling and analysis can be helpful in the FIGURE 7. Digital sensors provide information on activity and location,
diagnosis of eye diseases [121], [122], [123] and diseases together with emotional data, with the purpose of creating a background
picture of health (salivary in-mouth sensors) [161].
other than ophthalmology [124]. Research and analysis of
tears is a promising and interesting opportunity for the body (e.g., sweat on the skin). Therefore, continuous real-
diagnosis of neurological diseases [125], such as multiple time stress detection is cumbersome.
sclerosis [126], [127], Alzheimer’s dementia [128], and
Parkinson’s disease [129], [130]. Biomarkers such as stress 4) URINE
hormones have also been studied for stress detection and
Urine is a metabolic fluid found in humans and animals.
measurement [131]. In tears, the glucocorticoid hormone
Kidneys clean the blood from toxins and other elements that
cortisol, glucose [132], and many other proteins [133], such
must be removed and transformed into urine. Urine flows
as lysozyme, lactoferrin, secretory immunoglobulin A, serum
to the ureters and then to the urinary bladder; finally, it is
albumin, lipocalin, and lipophilin.
discarded through the urethra [165].
Wearable devices (e.g., lenses) have been developed
Urine is an attractive fluid for the prognosis and diagnosis
and applied for the detection of biomarkers in tears, with
of certain diseases [166], [167], [168], [169], [170], [171],
emphasis on stress measurements such as cortisol [134],
[172]. Thus, urine is usually tested for (a) biomarkers that
[135], lysozyme [136], and glucose [137], [138], [139]
are not normally present due to illness or injury and (b)
(mainly for diabetic purposes). There is a correlation between
the detection of normal biomarkers that are increased or
glucose and cortisol levels and the stress response (cortisol
decreased, including urinary incontinence, glucose measure-
increases sugars (glucose) in the bloodstream [140]).
ment (for diabetic patients), and oxidative stress biomarker
measurement [170] as a chronic disease, rather than (an acute)
stress response, which is not very convenient.
3) SALIVA Glucose is an attractive biomarker for stress detection
Saliva is an attractive fluid after blood serum for detecting and measurement in the urine using wearable sensors.
stress via biomarkers and the autonomic nervous system. To accomplish this, wearable devices designed and developed
Sympathetic stimulation can differentiate salivary flow. for people with diabetes have already been developed [169].
Some biomarkers detected in saliva include alpha-amylase, Urine and blood glucose levels are potentially corre-
uric acid (UA), glucose, lactate, phosphate, cortisol, and lated [173].
lysozyme [141]. Diurnal and seasonal variations in hormones
such as cortisol (normally higher in the morning and lower at 5) INTERSTITIAL FLUID (ISF)
night) must be considered [142], [143]. Interstitial fluid (ISF) surrounds cells in the body, constituting
Research, detection, and measurement of salivary 15–25% of body weight [174]. It is comparable to blood
biomarkers [144], [145], [146], especially α-amylase [147], plasma and functions on one hand as a fuel station that
[148], [149], [150], [151], cortisol [151], [152], [153], provides nutrition to cells and on the other hand, as a waste
lysozyme [154], prolactin [155], [156], and glucose [157], product removal mechanism from the cells [175].
[158], [159] have been conducted. ISF is considered unexplored for biomarker detection
α-Amylase, which is secreted by the salivary glands, is an and measurement [176]. Therapeutic and opioid (drug)
ideal biomarker for stress detection in saliva because it is monitoring using the ISF biomarker detection [177], [178],
controlled by the autonomic nervous system. The response of [179], [180], [181] is promising. ISF is certainly advancing in
amylase to stress is fast and immediate, unlike cortisol, which wearable glucose measurements, particularly in patients with
takes time (minutes later) [160]. diabetes (usually type I diabetes) [182], [183], [184]. Glucose
Biochemical sensors [162], [163], [164] are sometimes and cortisol are probable biomarkers of ISF that are related to
wearable [86], [88], [90], [91], [96] and can measure (or are stress responses and can be measured.
almost there) most of these biomarkers, especially α-amylase ISF is primarily accessed and sampled using micronee-
and cortisol, which are the so-called stress hormones’’. dles [185], [186], [187], [188], [189] and iontophoretic
Wearable sensors in the mouth (saliva) are impractical to extraction [190], [191], [192]. Both procedures were mini-
use and difficult to wear, as they are in other areas of the mally invasive [193]. Wearable devices are available to detect

VOLUME 12, 2024 38225


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

and measure glucose and cortisol levels in response to stress


stimuli [182], [183], [184], [191], [194].
The combination of biomarker detection and measurement
from sweat and ISF with one sensor or one device platform
is another interesting idea and is probably a promising
opportunity for biomarker measurements because both sweat
and ISF sampling devices are located on the skin [195].
ISF access and sampling is a minimally invasive method
compared with noninvasive methods, which is a serious
disadvantage that prohibits market acceptability.

C. BODY MOVEMENTS
The use of new-generation three-dimensional (3D)
accelerometers and fitted gyroscopes, for example, on a
smartwatch, is an important tool that can be used to evaluate
stress.
The motion must be evaluated during the stress measure-
FIGURE 8. Speech production method (from the lungs and vocal folds to
ments. However, different analytical methods are required the nasal and oral cavities) [202].
in certain situations. For example, social stress differs from
the stress induced by running, walking, or rowing. Wearable
devices such as accelerometers and gyroscopes can generally
detect body activity movements and sometimes the exact can be assessed as a result of stress [203]. However, the
activity that occurs. Thus, these data can be co-evaluated for opposite effect was observed. That is, wrong – poor posture
the final outcome. to induce more stress (among other things such as adversely
Stress, particularly social stress, can also induce tremors. affect pulmonary function [204], adverse musculoskeletal
Usually, stress increases the heart rate, blood pressure, and conditions, etc.) and posture correction to decrease stress on
breathing rate, and muscles are likely to be more tensed. This individuals [205].
could induce shaking and tremors. Various wearable devices are available for studying and
A ballistocardiogram (BCG) measures the ballistic forces evaluating posture (body and head) [206], [207]. Smart
generated by the heart (during systole), and is useful for clothes [208], [209], [210], [211], [212], vests [213],
measuring the pulse transit time (PTT), heart pulse rate, [214], patches [215], and watches (wrist) [216]. Some
and pulse wave analysis fusion (PWA) to assess arterial technologies and sensors used for posture measurement eval-
stiffness. Theoretically, (cuffless) blood pressure can be uation include motion sensors [217], accelerometers [218],
obtained. A microelectromechanical system (MEMS) tri- strain gauges [219], flexible – piezoelectric sensors [220],
axial accelerometer is used for this measurement, usually at [221], fiber-optic goniometers [222], inductive sensors [223],
the wrist, ear, and other body positions. ergonomic dosimeters [224], and inertial sensors [225].

D. SPEECH F. HAND TREMOR


Speaking is quite complex; it goes through several stages A shaking hand may be a stress indicator, among many others
until the voice gets out in a meaningful manner. Many human of course. Some of the causes that may induce or increase the
organs are involved in this procedure, such as the lungs, symptoms of hand tremors include Parkinson’s disease, alco-
pharynx-vocal cords, oral and nasal cavities, and brain [196]. hol withdrawal, multiple sclerosis, stress-anxiety, fatigue, and
Speech exhibits certain characteristics. Has a tone, it has high caffeine consumption [226], [227], [228].
a flow, it has a volume, it is sometimes melodically or in Hypoglycemia may also be involved in hand tremors,
a strict tone, it is shadily or happily, it is lively or tired? and wearable devices that can detect hand tremors (as
Researchers have found that reading and speaking (especially a warning) for this life-threatening condition have been
aloud) involves activating the autonomic nervous system (by developed [229]. Furthermore, Parkinson’s can be evaluated
requiring respiration) [197], [198]. using hand tremors [230], [231].
All the above-mentioned (organs and body parts) must be Thus, existing devices for measuring hand tremors for
functional, in good health, well-coordinated, especially by the other purposes, even for common health monitoring (e.g.,
brain, and work in harmony. Thus, the induced stress may smartwatches) with accelerometers, gyroscopes, adequate
cause significant changes in speech [199], [200], [201]. computing resources, internet connectivity via Wi-Fi, Blue-
tooth, 4G, or 5G, and the use of artificial intelligence (AI)
E. BODY POSTURE can measure hand tremors [232] and can also be co-evaluated
Posture is considered to be affected by stress. Changes in with other biosignals and/or biomarkers for stress detection –
posture, along with other signs (such as breathing rate) measurement.

38226 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

III. STRESS DETECTION TECHNOLOGIES biomedicine to examine the optical characteristics of


A. OPTICAL hemoglobin) to measure arterial blood oxygen saturation
Light, either in the visible or infrared spectrum, was used for and hemodynamics in general. It was later implemented and
biomarker measurements. applied to the human scalp by analyzing the reflection of
transmitted–emitted NIR light to human tissue [253].
1) PHOTOPLETHYSMOGRAPHY (PPG) Functional infrared spectroscopy (fNIRS) is a portable,
wearable NIRS technique for the functional monitoring
PPG is a noninvasive technology in which light is emitted
and imaging of human brain hemodynamics. This method
from low-consumption and high-efficiency LEDs to measure
is used to monitor and evaluate several problems and
the heart rate, frequency, and other parameters through reflec-
conditions at the brain level, including epilepsy, cognitive
tion or light absorption [233], [234], [235]. When a heartbeat
dysfunction, and anxiety [48], [243]. In addition to other
occurs, blood flow increases and differentiates the reflection
methods, such as fMRI, electroencephalography (EEG) can
or absorption of light. Using digital signal processing filtering
provide a fuller and more comprehensive view of a human’s
methods (to reduce noise and motion artifacts), the blood
brain situation [257].
volume pulse (BVP) can be obtained [236], [237], [238].
Heart rate (HR), heart rate variability (HRV), and respira-
tory rate can be measured with high accuracy by measuring 3) VISUAL OBSERVATION - IMAGING
the frequency or period of pulses per unit time [239], [240], Visual observation can reveal human health conditions and
[241]. In addition, blood oxygen saturation (SpO2) can be possible health problems. Cases such as iris of the eye,
measured using PPG [242]. BP is also expected that BP will eyelid opening and closing lapses, skin discoloration, or facial
be continuously measured via PPG [48], [243], [244] using cues can be detected using a camera (with the assistance of
various methods and signal combinations (bio-impedance, appropriate software) [245], [246], [247], [248].
pulse wave transit time (PWTT), AI, etc.) with relatively good Wearable devices, such as helmets equipped with a CCD
measurement accuracy or even medical precision [245]. or CMOS camera sensor, along with facial image tracking,
recognition, object detection mechanisms, and data analysis,
2) INFRARED (IR), NEAR IR (NIR) SPECTROSCOPY (NIRS), which are usually implemented using AI machine learning
AND FUNCTIONAL NIR SPECTROSCOPY (ML) systems [261], [262], [263], are possible.
Infrared (IR) light is electromagnetic radiation (EMR), which Imaging techniques include thermal imaging (TI), hyper-
is similar to human visible light. The wavelength of human spectral imaging (HSI), and broadband imaging. Almost
visible light ranges from 400 nm to 700 nm [246]. Thus, all of these have already been used by researchers for
IR was greater than 700 nm (between 780 nm and 1 mm). emotional–cognitive stress detection [264], [265], [266],
Near-infrared (NIR) light is called infrared light, next to [267], [268].
human visible light, and there are also the terms mid- and
far-infrared light. The NIR wavelength ranges from 0.78 µm B. ELECTRICAL
to 3 µm (or 780 nm to 3000 nm) [247]. In the human body, organs such as the heart or brain produce
Pulse oximetry is a noninvasive, portable, and wearable electrical signals that often reveal the state in terms of health,
method that uses infrared light, usually in thin parts of activity, emotions, and stress at the time of measurement or
the human body (such as the fingertip, earlobe, and wrist), for some time if the measurement has some duration. Some
to measure oxygen concentration or oxygen saturation. A red measurements that can be made using wearable devices are
LED (wavelength of approximately 650 nm) and an infrared as follows:
(NIR) LED (wavelength of approximately 950 nm) were
used. Blood saturated with oxygen absorbs more infrared
light and allows more red light, and vice versa (when it is 1) ELECTROENCEPHALOGRAPHY (EEG)
low oxygen-saturated). Therefore, blood-oxygen saturation is It is a graphical recording of the electrical activity of the
measured by absorption or non-absorption of red and NIR surface layer of the brain from the scalp [269]. Among other
light [248], [249], [250]. methods, EEG is usually used for the diagnosis of certain
Infrared thermography (IRT) and infrared thermometry illnesses or conditions, such as epilepsy, anesthesia, sleeping
(IRTM) are noncontact, portable, and wearable IR tem- disorders, encephalopathies, coma, dementia, trauma, and
perature measurement methods. Every object emits more death (brain death).
infrared (IR) light when heated. IR lenses and thermopile EEG records the following rhythms Alpha (8 Hz to 13 Hz),
sensors [251] are typically used to sense heat (via IR light). Beta (13 Hz to 30 Hz), Gamma (30 Hz to 100 Hz), Delta (0
Using this method, the skin temperatures of humans and Hz to 4 Hz), and Theta (4 Hz to 8 Hz), at various locations
animals can be measured [252]. The measurement distance of the head [270]. For example, beta waves are located in
is also a factor, and all devices have limitations. the frontal and parietal regions and respond in a state of
Near-infrared spectroscopy (NIRS) is a non-invasive alertness, active thinking, panic attack, and concentration as
method that employs near-infrared light (first used in well as neurotransmitters and hormones related to adrenaline,

VOLUME 12, 2024 38227


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

cortisol, and dopamine, which are active or related to stress (in microsiemens (µS)). The conductance level typically
responses to stress stimuli. varies, and the manner in which it varies provides valuable
Therefore, the EEG method, alone or in combination with conclusions [60], [294], [295].
other methods (such as HRV and EDA), has been used in
research and many scientific articles have been published
on the detection and quantification of stress responses [55],
[271], [272], [273], [274], [275], [276]. Recently, the idea
of using wearable EEG devices have become increasingly
attractive for the detection of stress responses in real life,
working environments (particularly trucks and buses), and
mental health management [51], [277], [278], [279], [280],
[281], [282], [283], [284]. Commercially available wearable
EEG devices are available [285], [286], [287].

2) ELECTROCARDIOGRAPHY (ECG)
FIGURE 10. Electrodermal Activity (EDA). Baseline conductance or (tonic)
It is a graphical representation (voltage/time) of the electrical skin conductance level (SCL) and (phasic) skin conductance response
activity of the heart. A normal ECG has a unique char- (SCR) [296].

acteristics (marks, curves, peaks, etc.) that have a specific


To obtain valuable information on stress detection, analysis
serial–temporal sequence and duration within certain time
of the conductance signal level is required. The skin
limits [288].
conductance baseline in a graphical representation of the
individual person and the measurement must be delineated
because of the variations. The baseline response curve is
smooth, and any sharp (rapid/sudden) bend in this curve
indicates autonomic nervous system arousal after an external
or internal stimulus [297], [298].
To detect and evaluate stress, it is necessary to distinguish
between the baseline (tonic) conductance and phasic conduc-
tance, which may be the result of a stressor, as shown in
Fig. 10. SCR is typically combined with other methods (e.g.,
EEG and ECG) for stress assessment [299], [300], [301].

4) ELECTROMYOGRAPHY (EMG)
FIGURE 9. Typical ECG graphical representation with all associated
valuable information [289]. EMG is used for the diagnosis and detection of abnormalities
in the muscles, neurons, or both. Electrical signals obtained
The interpretation of an electrocardiogram and its possible using electrodes from neuronal and muscle activities have
exclusions from a normal electrocardiogram can assist physi- been collected, analyzed, and evaluated [302], [303]. The
cians in diagnosing certain diseases (such as atrial fibrillation, trapezius muscles have been considered as stress predictors
atrial flutter, arrhythmias, tachycardia, and bradycardia) and in some studies [304], [305]. The sole EMG method, or more
drawing conclusions on someone’s health condition [290]. frequently complementary to other methods, such as EEG,
Measurements, such as heart rate (HR) and heart rate ECG, and HRV, is used in some cases to detect stress response
variability (HRV or R-R Interval), can be extracted relatively and activity [61], [306], [307].
easily from an ECG, even from non-medically educated
personnel. Both HR and HRV have been associated with 5) BIO-IMPEDANCE
stress detection and measurement in many studies [31], [291]. Bioimpedance refers to the response of living organisms to
Wearable devices, such as watches, have ECG capabilities, an externally applied electric current. The body impedes
whereas others appear to have medical accuracy [292], [293]. the electric current flow. For instance, the blood impedance
is lower, whereas the fat impedance is higher. Therefore,
3) SKIN CONDUCTANCE RESPONSE (SCR) – GALVANIC bioimpedance is used for body composition and blood flow
SKIN RESPONSE (GSR) analysis [308], [309].
Occasionally, the skin becomes an adequate electrical With electrical impedance plethysmography (EIS) and
conductor. This usually occurs when there is psychological other techniques, the volume of a cardiac blood pulse can be
or physiological arousal. This is typically measured by measured along with the pulse wave velocity (PWV), pulse
using two electrodes. A small electrical charge was applied transit time (PTT), or pulse wave transit time (PWTT) [309],
between the two endpoints of the electrodes on the skin. [310], [311]. Knowing all the above measurements, cuffless
Skin conductivity was observed, monitored, and/or recorded blood pressure estimation is also possible [312], [313].

38228 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

Wearable patches and watches [314], [315], [316] use


electrical bioimpedance to measure cardiac output, stroke
volume, and blood pressure. Bioimpedance and other
measurements of the human body can assist in indirect
measurement and evaluation of stress.

C. ELECTROMECHANICAL
Electromechanical sensors are related to mechanical energy
and response – stimulus – motion, which is converted to an
electrical signal at an analogous rate, usually measured in
volts. For example, slow motion, low voltage, and fast motion
result in a high voltage [317].
FIGURE 11. Schematic representation and components of typical
Some mechanical applications in which electromechanical (electrochemical) biosensors [338].
sensors are applied to the human body in a wearable form
include stress gauges, strain, motion-inertial, and pressure
sensors. They can be used for dynamic monitoring of Wearable noninvasive (or semi-invasive) biochemical
human posture [318], tremors, pulses, oscillations, vibrations, sensors can be used to apply/measure parameters from
and other physiological and motion parameters [319], the epidermis (sweat), intestinal fluid, tears, saliva, and
[320], [321], [322], [323], [324]. They are used by ath- urine. [337], [339], [340]. Data signals are transmitted
letes and patients and generally for health monitoring wirelessly using RFID, Bluetooth, and other wireless access
purposes. technologies for the monitoring devices.
Inertia electromechanical sensors [325] are flourishing
E. BIO-KINETIC
nowadays and are embedded in smartwatches and smart-
Another recently investigated method for stress detection
phones in the form of accelerometers and gyroscopes
is movement analysis using accelerometers and gyroscopes.
for motion detection, such as positioning, steps, sleeping
The movement of the hands (e.g., trembling), typing on a
monitoring, athletic activities, wellness, and healthcare in
computer (change in rhythm, pattern of typing), and (nervous)
general [326]. It is interesting to use these sensors for several
movement of the body in general can reveal the existence of
reasons:
stress.
Ballistocardiogram (BCG): measurement of ballistic
Table 1 presents some of the most commonly used
forces generated by the heart (during systole) acting on the
wearable devices for stress-response measurements.
entire body [327].
Seismocardiogram (SCG): Measurement of mechanical IV. SENSOR DATA NETWORKING
vibrations induced by the heart locally on the chest ( A. DATA FORM
not for forces acting on the entire body, such as BCG) In general, we can categorize the data from the various stress-
[326], [328], [329]. detection wearable devices into two categories, which are the
BCG and SCG are useful for measuring pulse transit following:
time (PTT) and heart pulse rate. With pulse wave analysis
fusion (PWA) to assess arterial stiffness) and with other 1) RAW DATA
complementary measurements, theoretically, a (cuffless) Raw or primary data are the original data that have not been
blood pressure can be obtained [330], [331], [332]. Wearable processed or modified. The raw data are altered so that they
MEMS triaxial accelerometers and gyroscopes are used for can be useful for a process. Wearable devices with sensors
these measurements, typically in the chest, wrists, ears, and store, transfer, and produce processed data even in real-time
other body areas. (e.g., processing and transferring ECG or PPG signal data).

2) SHAPED DATA
D. BIOCHEMICAL - ELECTROCHEMICAL It is data that has undergone some kind of processing,
Biochemical biomarkers are another way to identify, detect, merging, or juxtaposition (e.g., in tables and datasets) with
and measure metabolites (such as saliva and sweat), volatile the aim of making the data easier to read–to compare, more
organic compounds (VOCs), ions, and hormones (stress meaningful, and semantic. Furthermore, the data can be
and anti-stress hormones) related to stress [333], [334], graphically illustrated and statistically processed.
[335], [336].
A biochemical-electrochemical sensor can convert a B. DATA TRANSFER
chemical (or biological) reaction response to an electrical Data transfer and signal technologies from wearable devices
signal, providing the potential for personalized continuous and sensors can be divided into two categories: wireless and
monitoring and treatment related to stress [337]. wired technologies.

VOLUME 12, 2024 38229


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

TABLE 1. Wearable devices used for stress detection purposes.

38230 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

TABLE 1. (Continued.) Wearable devices used for stress detection purposes.

VOLUME 12, 2024 38231


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

1) WIRELESS stress-detection device, which can then store, analyze,


In recent years, wireless data signal transfer has rapidly transform (to shaped data), and display data.
developed and implemented. Various wireless transmission The following table (TABLE 1) presents a list of available
methods and technologies were selected and applied accord- devices – sensors that use almost all the methods and
ing to the requirements of each application. The following are technologies mentioned in this document-at various locations
typically used: on the human body. These devices are available for commer-
cial, medical, academic, and research purposes. Alone or in
a: Wi-Fi combination, they can or may be used exclusively or partially
Wi-Fi is a standardized data transfer protocol used in almost to monitor stress.
every home. It has a small and medium coverage range,
which is usually larger than that of Bluetooth, offers excellent V. DATA ANALYSIS METHODS
networking, and provides powerful data security. A major scientific trend concerning the analysis of data
collected from wearable systems to detect and evaluate stress
involves Artificial Intelligence (AI) and, more specifically,
b: BLUETOOTH
Machine Learning (ML) and Deep Learning (DL) (both
Bluetooth is a standardized data transfer protocol that is
subsets of AI). During the last decade, AI has mainly been
widely used and provides a relatively small-to-medium
used for data analysis and evaluation of all collected data in
coverage range.
such a manner that the presence of stress can be predicted and
managed accordingly.
c: ANT+
ANT + is a short-distance data-transfer protocol similar to A. ARTIFICIAL INTELLIGENCE (AI)
Bluetooth, and its data-transfer volume is relatively small. AI is a major chapter in information technology and has
many subsets such as ML and DL. Also, Fuzzy Logic (FL)
d: CELLULAR TECHNOLOGY is applied to AI (as a form of AI).
Cellular technologies (2G, 3G, 4G, and 5G) are well-known ML is the most frequently used mechanism for stress data
mobile interconnections and communication network tech- analysis [370], [371], [372], [373], [374], [375], [376], [377],
nologies. If a wearable device has the necessary equipment which continuously improves the outcome (evaluation),
and a SIM card, it can connect and transfer data through as long as more data are collected. The main ML methods
this network, thereby providing a wide coverage range. that exist are as follows:
5G networking technology provides better connectivity than • supervised learning (regression, classification)
previous generation cellular network technologies, a faster • unsupervised learning (clustering)
response, lower latency, and higher data flow. However, it has • reinforcement learning (q-learning)
a few limitations (it is available mainly in urban areas with a DL [274], [377], [378] imitates the human brain to learn
high intensity and frequency of 24GHz and above). and uses a large amount of data (usually more than a common
ML system). Neural Networks (NN) form the backbone of
e: NFC AND RFID DL systems.
Near-field communication (NFC) and radio frequency iden- Some ML algorithms used for the data analysis of a stress-
tification (RFID) are sets of protocols for establishing one- detection system are as follows:
or two-way communication between electronic devices at • Random Forest [379], [380], [381], [382]
distances of 5 cm or less for NFC and a maximum of 25 m to • Support Vector Machine (SVM) [383], [384], [385],
100 m (using batteries) for RFID. [386], [387]
• k-Nearest Neighbors (kNN) [388], [389], [390]
2) WIRED • Fuzzy Logic [391], [392], [393], [394]
Wireline data transfer offers no new or innovative solutions • Decision Trees [395], [396]
in the research field. They have existed for many years with • Principal Component Analysis (PCA) [388], [397],
some improvements over time. The volume of data from [398]
biometric sensors of this type is usually not large and the • Artificial Neural Network (ANN) [399], [400], [401]
transfer speed is not very demanding. Universal Serial Bus • Ensemble Methods [299], [402]
(USB) cables and various types of electrodes were used. • Naïve Bayes [385], [403]
Electrodes (and common cables) are electric conductors • Linear Discriminant Analysis [388], [404], [405]
that can carry electric signals from non-metallic surfaces,
such as solids (human skin, liquids, vacuum, gases, etc.). B. STATISTICAL ANALYSIS (SA) AND DATA MINING (DM)
Amorphous carbon, gold, and platinum are some of the SA and DM are important tools in modern ML [406].
materials that can be used in electrodes. Common electric Statistics are used in the data analysis of a stress-detection
cables can transfer raw signals from electrodes to a wearable mechanism, particularly at the pre-analysis level or as an

38232 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

TABLE 2. Summary of multimodal human stress detection studies (with and without wearables) [407].

additional support tool to extract meaningful relationships measurements. Modern ECG devices can calculate and
between the data. derive heart rate (HR) and heart rate variability (HRV-RR
The following table (TABLE 2) presents modalities – intervals). The galvanic skin response (GSR) is a signal
combined methods, along with models of artificial intel- with some useful characteristics: a quick response to stress
ligence (AI) algorithms and data classifications used in (using the sympathetic system) and no intervention on the
various studies, and the accuracy achieved in each case from skin from the parasympathetic system (sympathetic and
2010 onwards Arsalan et al. [407]. parasympathetic are antagonistic systems). Therefore, this is
a less complicated signal for stress detection. A combination
VI. DISCUSSION of HR-HRV and GSR signals is recommended and wearable
A. STRESS DETECTION MAIN METHODS AND devices are available for (this) implementation.
ADDITIONAL PROPOSALS New devices (smartwatches) that can measure dynami-
Looking at the present and both tables (TABLE 1 and cally, cuffless, and continuous blood pressure using bio-
TABLE 2, devices and methods of stress detection), we can impedance and other methods (pulse transit time [PTT])
draw some conclusions and propose solutions. are promising innovative solutions for stress measurement
optimization.
1) AUTONOMIC NERVOUS SYSTEM (ANS) Brain activity measurement techniques such as electroen-
RESPONSE-RELATED MEASUREMENTS FOR STRESS cephalography (EEG) are regarded as impractical; however,
Electrocardiography (ECG) is the most widely used in some cases (e.g., offices and cockpits), when a cask,
technique to acquire characteristic signals for stress helmet, or headset can be worn, they are efficient. Usually,

VOLUME 12, 2024 38233


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

the HR and other signals (such as HRV and temperature) can Real-life continuous stress measurement scenarios using
also be measured using EEG devices. wearable devices are significantly more complex. A more
sophisticated and multi-input data approach from (both)
2) HYPOTHALAMUS-PITUITARY-ADRENAL (HPA) AXIS the ANS and HPA axis and other feedback is required.
RESPONSE-RELATED STRESS MEASUREMENTS Verification of the measurements in experiments is often
Cortisol is a well-known stress hormone. A wearable performed using hormone measurements such as cortisol,
device with the ability to measure this hormone or other enzymes such as α-amylase (in blood, saliva, and sweat), and
related (to stress) hormones, enzymes, and proteins is also questionnaires such as the Perceived Stress Scale (PSS) and
an innovative, complementary, and auditable method for its derivatives.
detecting, calibrating, and quantifying stress. Watches are the most widely used wearable device. There
are smartwatches (TABLE 1) for commercial, medical, and
3) ADDITIONAL MEASUREMENTS FOR STRESS research uses that can or are about to measure the following
Well-established devices, methods, and technologies in areas continuously or frequently:
other than stress are presented here for the sole, but
1. Heart rate (HR), heart rate variability (HRV), and blood
mainly complementary, use of stress detection and stress
oxygen saturation (SpO2) via photoplethysmography
monitoring optimization with all the appropriate adaptations.
(PPG optical sensor).
For example, wearable devices for Parkinson’s disease (hand
2. Ambient and skin temperature (ST) measurements using
tremor), diabetes (continuous glucose measurement), driver
infrared (IR) sensors.
and pilot tracking of alertness (eye dynamics), and other
3. Movement: steps – physical activity tracking via 3-axis
devices or sensors for athletic performance monitoring have
(3D) accelerometer (ACC) and gyroscope.
been correlated with stress-response detection and grading.
4. Electrodermal activity (EDA) via skin conductance
response (SCR) sensor.
B. STRESS IDENTIFICATION, CATEGORIZATION, AND
5. Cuffless blood pressure (BP) via a bioimpedance sensor
PARADIGMS
or pulse transit time (PTT) estimation and PPG sensor
This review can also help better understand stress-cause
synergy.
identification (at least on some occasions), stress categoriza-
6. Cortisol measurements using sweat biochemical sensors.
tion (positive or negative stress, exogenous or endogenous
7. Glucose via PPG sensor or similar sensors
stress), and stress-mimicking (pseudo-stress) recognition.
The following are some characteristic paradigms: In the current decade, smartwatches with all the above
1) A stress response was detected and a 3D accelerometer features (sensors) are not unlikely to be released. Further
detected intense physical activity. This was probably research, improvements, financial expenditure, and firm
because of physical activity, which is a positive endoge- decisions are required.
nous stressor. Collection, correlation, and data analysis in real time and
2) A stress response was detected and the outside tem- continually (removing noise and artifacts by constantly reset-
perature changed significantly. Causes of exogenous ting the baseline of fluctuating values) using multiple sensors
temperature. and devices may be challenging. For instance, hormones
3) A stress response was detected and the skin and body such as cortisol fluctuate throughout the day. Therefore,
temperatures were high. This was probably due to fever determining the baseline measurement of individual and
(endogenous–negative stress). contextual factors with unknown stressors anytime and
In the above paradigms, if cortisol levels do not increase anywhere is an arduous task.
(after a few minutes), then it is not stress (pseudo-stress). Introducing diversity in hardware by combining different
Furthermore, artificial intelligence (AI) with self-learning sensors can significantly increase the complexity of the
or guided learning methods in combination with the afore- hardware, data, and software. However, adding new hardware
mentioned methods can prove very helpful in this area. to an already tested solution or changing the dataset may not
Additionally, searching for a source of stress through self- guarantee a better performance or prevent it from worsening.
awareness may be helpful. Therefore, hardware must be set with appropriate software
to achieve optimal performance. Hence, further research and
C. SUGGESTIONS, COMPLEXITY, COST, AND experimentation on hardware and software correlations is
LEGAL-ETHICAL ISSUES required.
Most scientific studies have focused on measuring the acute The added cost is certain to develop and produce; for
stress caused by specific stressors. Researchers established example, a smartwatch that combines most of the aforemen-
a baseline measurement (usually using reliable and trial tioned measurements. However, this is a shared cost because
methods) before the experiment. They then measured stress the end-user will benefit not only from stress monitoring
levels during the initiation of the stressor, its peak, and during but also from many other important measurements, such as
the recovery period. Finally, the results were compared to glucose (very significant for diabetic or pro-diabetic people),
estimate the precision of the trial (method). blood pressure (for people with BP problems), and health and

38234 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

athletic performance monitoring. Technological enthusiasts [3] (2016). Investing in Treatment for Depression and Anxiety Leads
are interested in these products. to Fourfold Return. Accessed: Aug. 9, 2022. [Online]. Available:
https://www.who.int/news/item/13-04-2016-investing-in-treatment-for-
Legal-ethical issues: Data from wearable devices or depression-and-anxiety-leads-to-fourfold-return
sensors for measuring stress may expose personal data of [4] W. B. Cannon, Bodily Changes in Pain, Hunger, Fear and Rage:
health interest. In other words, they may fall into medical An Account of Recent Researches into The Function of Emotional
Excitement. New York, NY, USA: D Appleton & Company, 1915,
confidentiality. A cardiogram can reveal a disease (e.g., pp. 13–311, doi: 10.1037/10013-000.
atrial fibrillation or heart attack). EEG can also reveal [5] T. W. Colligan and E. M. Higgins, ‘‘Workplace stress: Etiology and
brain dysfunction (e.g., epilepsy). Therefore, during research, consequences,’’ J. Workplace Behav. Health, vol. 21, no. 2, pp. 89–97,
Jul. 2006, doi: 10.1300/j490v21n02_07.
especially when these systems are applied to a large number [6] E. Papaefstathiou, A. Apostolopoulou, E. Papaefstathiou, K. Moysidis,
of people, the consent of the subjects should be required, the K. Hatzimouratidis, and P. Sarafis, ‘‘The impact of burnout and occu-
privacy of the data should be ensured, and they should be used pational stress on sexual function in both male and female individuals:
A cross-sectional study,’’ Int. J. Impotence Res., vol. 32, no. 5,
only for stress measurements (following the 1964 Helsinki pp. 510–519, Sep. 2020, doi: 10.1038/s41443-019-0170-7.
Declaration and its later amendments or comparable ethical [7] Harvard Health. Understanding the Stress Response. Accessed:
standards). In addition, attention should be paid to local Aug. 11, 2022. [Online]. Available: https://www.health.harvard.
edu/staying-healthy/understanding-the-stress-response
(region-country) legislation and its particularities. [8] The Truth About Lie Detectors (Aka Polygraph Tests). Accessed:
Oct. 25, 2023. [Online]. Available: https://www.apa.org/topics/cognitive-
neuroscience/polygraph
VII. CONCLUSION [9] S. Cohen, T. Kamarck, and R. Mermelstein, ‘‘A global measure of
Stress is a reaction to events or conditions that can be perceived stress,’’ J. Health Social Behav., vol. 24, no. 4, p. 385,
referred to as a stressor. This seriously affects daily life and Dec. 1983, doi: 10.2307/2136404.
[10] E.-H. Lee, ‘‘Review of the psychometric evidence of the perceived stress
has a significant impact on human wellbeing. In this study, scale,’’ Asian Nursing Res., vol. 6, no. 4, pp. 121–127, Dec. 2012, doi:
the most important technological methods using wearable 10.1016/j.anr.2012.08.004.
sensors are presented for commercially available products or [11] A. K. W. Chee, ‘‘Quantitative dopant profiling by energy filtering in
the scanning electron microscope,’’ IEEE Trans. Device Mater. Rel.,
laboratory/scientific products. Many commercially available vol. 16, no. 2, pp. 138–148, Jun. 2016, doi: 10.1109/TDMR.2016.
user-friendly wearable devices and sensors provide large 2531506.
datasets for analysis, stress detection, and stress management. [12] D. Solanki, G. Cay, M. Rumon, V. Ravichandran, and K. Mankodiya,
‘‘A step towards design and validation of a wearable multi-sensory
This review helped to identify new research areas. smart-textile system for respiration monitoring,’’ in Proc. IEEE
Measurement of stress, especially using both the autonomic Sensors, May 2022, pp. 1–4, doi: 10.1109/SENSORS52175.2022.
nervous system (ANS) and hypothalamic–pituitary–adrenal 9967050.
[13] P. Corbishley and E. Rodriguez-Villegas, ‘‘Breathing detection: Towards
(HPA) axis responses, can theoretically and potentially offer a miniaturized, wearable, battery-operated monitoring system,’’ IEEE
stress detection and pseudo-stress detection optimization. Trans. Biomed. Eng., vol. 55, no. 1, pp. 196–204, Jan. 2008, doi:
Data-driven AI involves feeding more data (the more, the 10.1109/TBME.2007.910679.
[14] S. M. Mirvakili, K. Broderick, and R. S. Langer, ‘‘A new approach for
better, and up to a certain limit). Data from the distributed microfabrication of printed circuit boards with ultrafine traces,’’ ACS
sensors were also more abundant and of better quality. Appl. Mater. Interface, vol. 11, no. 38, pp. 35376–35381, Sep. 2019, doi:
Little to no research has been conducted on the HPA axis 10.1021/acsami.9b08761.
[15] J. Chen, M. Abbod, and J.-S. Shieh, ‘‘Pain and stress detection using
and other physical secretions using wearables. Numerous wearable sensors and devices—A review,’’ Sensors, vol. 21, no. 4,
scenarios and combinations (at both the hardware and p. 1030, Feb. 2021, doi: 10.3390/s21041030.
software data analysis levels) have been developed and [16] Stress Effects on the Body. Accessed: Aug. 5, 2022. [Online]. Available:
https://www.apa.org/topics/stress/body
researched. Generally, a broader view of the topic is offered [17] E. Won and Y.-K. Kim, ‘‘Stress, the autonomic nervous system,
and diversity is encouraged if complexity is welcomed, and the immune-kynurenine pathway in the etiology of depression,’’
practically feasible, and economically attractive. Current Neuropharmacol., vol. 14, no. 7, pp. 665–673, Aug. 2016, doi:
10.2174/1570159x14666151208113006.
The ultimate aim of a stress detection system is to [18] H. Lin, H. Lin, W. Lin, and A. C. Huang, ‘‘Effects of stress, depression,
accurately detect stress, offer a cost-effective commercial and their interaction on heart rate, skin conductance, finger temperature,
solution, and operate under daily life conditions without and respiratory rate: Sympathetic-parasympathetic hypothesis of stress
and depression,’’ J. Clin. Psychol., vol. 67, no. 10, pp. 1080–1091,
negatively affecting the individual’s life. This review will Oct. 2011, doi: 10.1002/jclp.20833.
assist researchers and scientists in selecting the optimum [19] A. Mohammadi, M. Fakharzadeh, and B. Baraeinejad, ‘‘An inte-
stress detection method for each daily life scenario, together grated human stress detection sensor using supervised algorithms,’’
IEEE Sensors J., vol. 22, no. 8, pp. 8216–8223, Apr. 2022, doi:
with the appropriate data analysis method that should be used.
10.1109/JSEN.2022.3157795.
The ultimate goals of stress detection and management can be [20] G. Giannakakis, D. Grigoriadis, K. Giannakaki, O. Simantiraki, A. Roni-
achieved by combining this information. otis, and M. Tsiknakis, ‘‘Review on psychological stress detection using
biosignals,’’ IEEE Trans. Affect. Comput., vol. 13, no. 1, pp. 440–460,
Jan. 2022, doi: 10.1109/TAFFC.2019.2927337.
REFERENCES [21] Cleveland Clinic. Heart Rate Variability (HRV): What it is and
How You Can Track it. Accessed: Jul. 18, 2022. [Online]. Available:
[1] M. Kalia, ‘‘Assessing the economic impact of stress—The modern day https://my.clevelandclinic.org/health/symptoms/21773-heart-rate-
hidden epidemic,’’ Metabolism, vol. 51, no. 6, pp. 49–53, Jun. 2002, doi: variability-hrv
10.1053/meta.2002.33193. [22] F. Bahmed, F. Khatoon, and B. R. Reddy, ‘‘Relation between respiratory
[2] D. Boyd. Workplace Stress. Amer. Inst. Stress. Accessed: Aug. 9, 2022. rate and heart rate—A comparative study,’’ Indian J. Clin. Anatomy
[Online]. Available: https://www.stress.org/workplace-stress Physiol., vol. 3, no. 4, p. 436, 2016.

VOLUME 12, 2024 38235


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[23] M. Richter and R. A. Wright, ‘‘Autonomic nervous system (ANS),’’ [39] S. D. Aeschliman, M. S. Blue, K. B. Williams, C. M. Cobb, and
in Encyclopedia of Behavioral Medicine (ANS), M. D. Gellman and S. R. MacNeill, ‘‘A preliminary study on oxygen saturation levels of
J. R. Turner, Eds. New York, NY, USA: Springer, 2013, pp. 165–168, doi: patients during periodontal surgery with and without oral conscious seda-
10.1007/978-1-4419-1005-9_790. tion using diazepam,’’ J. Periodontology, vol. 74, no. 7, pp. 1056–1059,
[24] P. Karthikeyan, M. Murugappan, and S. Yaacob, ‘‘Detection Jul. 2003, doi: 10.1902/jop.2003.74.7.1056.
of human stress using short-term ECG and HRV signals,’’ [40] T. Tamura, ‘‘Current progress of photoplethysmography and SPO2
J. Mech. Med. Biol., vol. 13, no. 2, Apr. 2013, Art. no. 1350038, for health monitoring,’’ Biomed. Eng. Lett., vol. 9, no. 1, pp. 21–36,
doi: 10.1142/s0219519413500383. Feb. 2019, doi: 10.1007/s13534-019-00097-w.
[25] E. J. Pino, J. A. P. Chávez, and P. Aqueveque, ‘‘BCG algorithm [41] Z. Zhang and R. Khatami, ‘‘Can we trust the oxygen saturation measured
for unobtrusive heart rate monitoring,’’ in Proc. IEEE Healthcare by consumer smartwatches?’’ Lancet Respiratory Med., vol. 10, no. 5,
Innov. Point Care Technol. (HI-POCT), Nov. 2017, pp. 180–183, doi: pp. e47–e48, May 2022, doi: 10.1016/s2213-2600(22)00103-5.
10.1109/HIC.2017.8227614. [42] A. M. Gordon and W. B. Mendes, ‘‘A large-scale study of stress,
[26] O. T. Inan, P.-F. Migeotte, K.-S. Park, M. Etemadi, K. Tavakolian, emotions, and blood pressure in daily life using a digital platform,’’
R. Casanella, J. Zanetti, J. Tank, I. Funtova, G. K. Prisk, and M. Di Proc. Nat. Acad. Sci. USA, vol. 118, no. 31, pp. 1–12, Aug. 2021, doi:
Rienzo, ‘‘Ballistocardiography and seismocardiography: A review of 10.1073/pnas.2105573118.
recent advances,’’ IEEE J. Biomed. Health Informat., vol. 19, no. 4, [43] N. M. Kaplan, ‘‘Stress, the sympathetic nervous system and hyper-
pp. 1414–1427, Jul. 2015, doi: 10.1109/JBHI.2014.2361732. tension,’’ J. Hum. Stress, vol. 4, no. 3, pp. 29–34, Sep. 1978, doi:
[27] M. Haescher, D. J. C. Matthies, J. Trimpop, and B. Urban, ‘‘A study 10.1080/0097840x.1978.9934993.
on measuring heart- and respiration-rate via Wrist–Worn accelerometer- [44] D. Konstantinidis, P. Iliakis, F. Tatakis, K. Thomopoulos, K. Dimitriadis,
based seismocardiography (SCG) in comparison to commonly applied D. Tousoulis, and K. Tsioufis, ‘‘Wearable blood pressure measurement
technologies,’’ in Proc. 2nd Int. Workshop Sensor-Activity Recog- devices and new approaches in hypertension management: The digital
nit. Interact. New York, NY, USA: Association for Computing Machin- era,’’ J. Hum. Hypertension, vol. 36, no. 11, pp. 945–951, Mar. 2022, doi:
ery, Jun. 2015, pp. 1–6, doi: 10.1145/2790044.2790054. 10.1038/s41371-022-00675-z.
[28] M. Swapna, U. M. Viswanadhula, R. Aluvalu, V. Vardharajan, and [45] M. Falter, M. Scherrenberg, K. Driesen, Z. Pieters, T. Kaihara, L. Xu,
K. Kotecha, ‘‘Bio-signals in medical applications and challenges using E. G. Caiani, P. Castiglioni, A. Faini, G. Parati, and P. Dendale,
artificial intelligence,’’ J. Sensor Actuator Netw., vol. 11, no. 1, p. 17, ‘‘Smartwatch-based blood pressure measurement demonstrates insuf-
Feb. 2022, doi: 10.3390/jsan11010017. ficient accuracy,’’ Frontiers Cardiovascular Med., vol. 9, pp. 1–10,
[29] M. S. Bin, O. O. Khalifa, and R. A. Saeed, ‘‘Real-time personalized Jul. 2022.
stress detection from physiological signals,’’ in Proc. Int. Conf. Comput., [46] T. G. Pickering, ‘‘Mental stress as a causal factor in the development
Control, Netw., Electron. Embedded Syst. Eng. (ICCNEEE), Sep. 2015, of hypertension and cardiovascular disease,’’ Current Hypertension Rep.,
pp. 28–29. vol. 3, no. 3, pp. 249–254, Jun. 2001, doi: 10.1007/s11906-001-0047-1.
[30] S. Yoon, J. K. Sim, and Y.-H. Cho, ‘‘A flexible and wearable human [47] C.-S. Kim, A. M. Carek, O. T. Inan, R. Mukkamala, and J.-O. Hahn,
stress monitoring patch,’’ Sci. Rep., vol. 6, no. 1, pp. 1–52, Mar. 2016, ‘‘Ballistocardiogram-based approach to cuffless blood pressure
doi: 10.1038/srep23468. monitoring: Proof of concept and potential challenges,’’ IEEE
[31] K. Kyriakou, B. Resch, G. Sagl, A. Petutschnig, C. Werner, D. Niederseer, Trans. Biomed. Eng., vol. 65, no. 11, pp. 2384–2391, Nov. 2018,
M. Liedlgruber, F. Wilhelm, T. Osborne, and J. Pykett, ‘‘Detect- doi: 10.1109/TBME.2018.2797239.
ing moments of stress from measurements of wearable physiolog- [48] B. Ibrahim and R. Jafari, ‘‘Cuffless blood pressure monitoring from a
ical sensors,’’ Sensors, vol. 19, no. 17, p. 3805, Sep. 2019, doi: wristband with calibration-free algorithms for sensing location based on
10.3390/s19173805. bio-impedance sensor array and autoencoder,’’ Sci. Rep., vol. 12, no. 1,
[32] K. A. Herborn, J. L. Graves, P. Jerem, N. P. Evans, R. Nager, pp. 1–3, Jan. 2022, doi: 10.1038/s41598-021-03612-1.
D. J. McCafferty, and D. E. F. McKeegan, ‘‘Skin temperature reveals [49] OMRON Healthcare. Blood Pressure Watch | OMRON HeartGuide
the intensity of acute stress,’’ Physiol. Behav., vol. 152, pp. 225–230, Wrist BP Monitor. Accessed: Mar. 20, 2023.[Online]. Available:
Dec. 2015, doi: 10.1016/j.physbeh.2015.09.032. https://omronhealthcare.com/products/heartguide-wearable-blood-
[33] V. Chandra, A. Priyarup, and D. Sethia, ‘‘Comparative study of physi- pressure-monitor-bp8000m/
ological signals from Empatica E4 wristband for stress classification,’’ [50] P. Seoane-Collazo, J. Fernø, F. Gonzalez, C. Diéguez, R. Leis,
in Advances in Computing and Data Sciences (Communications in R. Nogueiras, and M. López, ‘‘Hypothalamic-autonomic control of
Computer and Information Science), M. Singh, V. Tyagi, P. K. Gupta, energy homeostasis,’’ Endocrine, vol. 50, no. 2, pp. 276–291, Nov. 2015,
J. Flusser, and V. R. Sonawane, Eds. Cham, Switzerland: Springer, 2021, doi: 10.1007/s12020-015-0658-y.
pp. 218–229, doi: 10.1007/978-3-030-88244-0_21. [51] J. W. Ahn, Y. Ku, and H. C. Kim, ‘‘A novel wearable EEG and ECG
[34] A. Kamišalić, I. Fister, M. Turkanović, and S. Karakatič, ‘‘Sen- recording system for stress assessment,’’ Sensors, vol. 19, no. 9, p. 1991,
sors and functionalities of non-invasive wrist-wearable devices: A Apr. 2019, doi: 10.3390/s19091991.
review,’’ Sensors, vol. 18, no. 6, p. 1714, May 2018, doi: 10.3390/ [52] E. Gratton, V. Toronov, U. Wolf, M. Wolf, and A. Webb, ‘‘Measurement
s18061714. of brain activity by near-infrared light,’’ J. Biomed. Opt., vol. 10, no. 1,
[35] H.-Y. Chen, A. Chen, and C. Chen, ‘‘Investigation of the impact of Feb. 2005, Art. no. 011008, doi: 10.1117/1.1854673.
infrared sensors on core body temperature monitoring by comparing [53] P. Pinti, C. Aichelburg, S. Gilbert, A. Hamilton, J. Hirsch, P. Burgess, and
measurement sites,’’ Sensors, vol. 20, no. 10, p. 2885, May 2020, doi: I. Tachtsidis, ‘‘A review on the use of wearable functional near-infrared
10.3390/s20102885. spectroscopy in naturalistic environments,’’ Jpn. Psychol. Res., vol. 60,
[36] M. Santos, S. Vollam, M. A. Pimentel, C. Areia, L. Young, C. Roman, no. 4, pp. 347–373, Oct. 2018, doi: 10.1111/jpr.12206.
J. Ede, P. Piper, E. King, M. Harford, A. Shah, O. Gustafson, [54] M. Haak, S. Bos, S. Panic, and L. Rothkrantz, ‘‘Detecting stress using
L. Tarassenko, and P. Watkinson, ‘‘The use of wearable pulse oximeters eye blinks during game playing,’’ presented at the Eur. GAME-ON Conf.,
in the prompt detection of hypoxemia and during movement: Diagnostic 2009.
accuracy study,’’ J. Med. Internet Res., vol. 24, no. 2, Feb. 2022, [55] X. Hou, Y. Liu, O. Sourina, Y. R. E. Tan, L. Wang, and W. Mueller-Wittig,
Art. no. e28890, doi: 10.2196/28890. ‘‘EEG based stress monitoring,’’ in Proc. IEEE Int. Conf. Syst., Man,
[37] O. Yossef Hay, M. Cohen, I. Nitzan, Y. Kasirer, S. Shahroor-Karni, Cybern., Oct. 2015, pp. 3110–3115, doi: 10.1109/SMC.2015.540.
Y. Yitzhaky, S. Engelberg, and M. Nitzan, ‘‘Pulse oximetry with [56] E. T. Attar, V. Balasubramanian, E. Subasi, and M. Kaya, ‘‘Stress analysis
two infrared wavelengths without calibration in extracted arterial based on simultaneous heart rate variability and EEG monitoring,’’
blood,’’ Sensors, vol. 18, no. 10, p. 3457, Oct. 2018, doi: 10.3390/ IEEE J. Translational Eng. Health Med., vol. 9, pp. 1–7, 2021, doi:
s18103457. 10.1109/JTEHM.2021.3106803.
[38] D. Nachman, Y. Gepner, N. Goldstein, E. Kabakov, A. B. Ishay, [57] H. D. Critchley, ‘‘Review: Electrodermal responses: What happens in
R. Littman, Y. Azmon, E. Jaffe, and A. Eisenkraft, ‘‘Comparing blood the brain,’’ Neuroscientist, vol. 8, no. 2, pp. 132–142, Apr. 2002, doi:
pressure measurements between a photoplethysmography-based and a 10.1177/107385840200800209.
standard cuff-based manometry device,’’ Sci. Rep., vol. 10, no. 1, [58] W. Boucsein, Electrodermal Activity. Boston, MA, USA: Springer, 2012,
p. 16116, Sep. 2020, doi: 10.1038/s41598-020-73172-3. doi: 10.1007/978-1-4614-1126-0.

38236 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[59] C. Tronstad, H. Kalvøy, S. Grimnes, and Ø. G. Martinsen, ‘‘Waveform [78] G. Giannakakis, M. Pediaditis, D. Manousos, E. Kazantzaki, F. Chiarugi,
difference between skin conductance and skin potential responses in rela- P. G. Simos, K. Marias, and M. Tsiknakis, ‘‘Stress and anxiety detection
tion to electrical and evaporative properties of skin,’’ Psychophysiology, using facial cues from videos,’’ Biomed. Signal Process. Control, vol. 31,
vol. 50, no. 11, pp. 1070–1078, Nov. 2013, doi: 10.1111/psyp.12092. pp. 89–101, Jan. 2017, doi: 10.1016/j.bspc.2016.06.020.
[60] M. Benedek and C. Kaernbach, ‘‘A continuous measure of phasic [79] Tobii Pro Glasses 3 | Latest in Wearable Eye Tracking. Accessed:
electrodermal activity,’’ J. Neurosci. Methods, vol. 190, no. 1, pp. 80–91, Mar. 20, 2023. [Online]. Available: https://www.tobii.com/products/eye-
Jun. 2010, doi: 10.1016/j.jneumeth.2010.04.028. trackers/wearables/tobii-pro-glasses-3
[61] S. Pourmohammadi and A. Maleki, ‘‘Stress detection using ECG [80] F. Dehais, M. Causse, and J. Pastor, ‘‘Embedded eye tracker in a
and EMG signals: A comprehensive study,’’ Comput. Methods real aircraft: New perspectives on pilot/aircraft interaction monitoring,’’
Programs Biomed., vol. 193, Sep. 2020, Art. no. 105482, doi: presented at the Proc. presented at the ICRAT, 2008, pp. 1–7. Accessed:
10.1016/j.cmpb.2020.105482. Mar. 9, 2023. [Online]. Available: http://www.icrat.org/
[62] Y. Masaoka, A. Kanamaru, and I. Homma, ‘‘Anxiety and respiration,’’ [81] S. M. Smith and W. W. Vale, ‘‘The role of the hypothalamic-
in Respiration and Emotion, Y. Haruki, I. Homma, A. Umezawa, pituitary-adrenal axis in neuroendocrine responses to stress,’’ Dia-
and Y. Masaoka, Eds. Tokyo, Japan: Springer, 2001, pp. 55–64, doi: logues Clin. Neurosci., vol. 8, no. 4, pp. 383–395, Dec. 2006, doi:
10.1007/978-4-431-67901-1_6. 10.31887/dcns.2006.8.4/ssmith.
[63] E. Piuzzi, S. Pisa, E. Pittella, L. Podestà, and S. Sangiovanni, ‘‘Wearable [82] J. A. Sheng, N. J. Bales, S. A. Myers, A. I. Bautista, M. Roueinfar,
belt with built-in textile electrodes for cardio—Respiratory monitoring,’’ T. M. Hale, and R. J. Handa, ‘‘The hypothalamic-pituitary-adrenal axis:
Sensors, vol. 20, no. 16, p. 4500, Aug. 2020, doi: 10.3390/s20164500. Development, programming actions of hormones, and maternal-fetal
[64] V. Jeyhani, T. Vuorinen, M. Mäntysalo, and A. Vehkaoja, ‘‘Comparison interactions,’’ Frontiers Behav. Neurosci., vol. 14, pp. 1–10, Jan. 2021.
of simple algorithms for estimating respiration rate from electrical [83] Electrochemical Sensor—An Overview | ScienceDirect Topics. Accessed:
impedance pneumography signals in wearable devices,’’ Health Technol., Aug. 11, 2022. [Online]. Available: https://www.sciencedirect.
vol. 7, no. 1, pp. 21–31, Dec. 2016, doi: 10.1007/s12553-016-0156-0. com/topics/chemistry/electrochemical-sensor
[65] P. Sharma, X. Hui, J. Zhou, T. B. Conroy, and E. C. Kan, ‘‘Wearable radio- [84] O. Parlak, ‘‘Portable and wearable real-time stress monitoring: A critical
frequency sensing of respiratory rate, respiratory volume, and heart rate,’’ review,’’ Sensors Actuators Rep., vol. 3, Nov. 2021, Art. no. 100036, doi:
NPJ Digit. Med., vol. 3, no. 1, pp. 1–15, Jul. 2020, doi: 10.1038/s41746- 10.1016/j.snr.2021.100036.
020-0307-6. [85] H. Teymourian, M. Parrilla, J. R. Sempionatto, N. F. Montiel,
[66] A. Panahi, A. Hassanzadeh, and A. Moulavi, ‘‘Design of a low A. Barfidokht, R. Van Echelpoel, K. De Wael, and J. Wang, ‘‘Wearable
cost, double triangle, piezoelectric sensor for respiratory monitoring electrochemical sensors for the monitoring and screening of drugs,’’ ACS
applications,’’ Sens. Bio-Sens. Res., vol. 30, Dec. 2020, Art. no. 100378, Sensors, vol. 5, no. 9, pp. 2679–2700, Sep. 2020, doi: 10.1021/acssen-
doi: 10.1016/j.sbsr.2020.100378. sors.0c01318.
[86] J. Min, J. R. Sempionatto, H. Teymourian, J. Wang, and W. Gao,
[67] J. W. Jeong, Y. W. Jang, I. Lee, S. Shin, and S. Kim, ‘‘Wearable respiratory
‘‘Wearable electrochemical biosensors in North America,’’
rate monitoring using piezo-resistive fabric sensor,’’ in Proc. World
Biosensors Bioelectron., vol. 172, Jan. 2021, Art. no. 112750, doi:
Congr. Med. Phys. Biomed. Eng., 2009, Munich, Germany. Berlin,
10.1016/j.bios.2020.112750.
Heidelberg: Springer, 2009, pp. 282–284, doi: 10.1007/978-3-642-
03904-1_78. [87] A. J. Bandodkar and J. Wang, ‘‘Non-invasive wearable electrochemical
sensors: A review,’’ Trends Biotechnol., vol. 32, no. 7, pp. 363–371,
[68] M. Chu, T. Nguyen, V. Pandey, Y. Zhou, H. N. Pham, R. Bar-Yoseph,
Jul. 2014.
S. Radom-Aizik, R. Jain, D. M. Cooper, and M. Khine, ‘‘Respiration rate
[88] K. J. Fraser, V. F. Curto, and S. Coyle, ‘‘Wearable electrochemical sensors
and volume measurements using wearable strain sensors,’’ NPJ Digit.
for monitoring performance athletes,’’ Proc. SPIE, vol. 8118, pp. 43–54,
Med., vol. 2, no. 1, pp. 1–20, Feb. 2019, doi: 10.1038/s41746-019-0083-
Sep. 2011, doi: 10.1117/12.895109.
3.
[89] T. Alves, P. Deroco, D. Wachholz Junior, L. Vidotto, and
[69] M. J. Tipton, A. Harper, J. F. R. Paton, and J. T. Costello, ‘‘The human
L. Kubota, ‘‘Wireless wearable electrochemical sensors: A review,’’
ventilatory response to stress: Rate or depth?’’ J. Physiol., vol. 595, no. 17,
Brazilian J. Anal. Chem., vol. 8, no. 10, pp. 1–12, Mar. 2021, doi:
pp. 5729–5752, Sep. 2017, doi: 10.1113/jp274596.
10.30744/brjac.2179-3425.rv-62-2020.
[70] A. Zaccaro, A. Piarulli, M. Laurino, E. Garbella, D. Menicucci,
[90] P. Swetha, U. Balijapalli, and S.-P. Feng, ‘‘Wireless accessing of salivary
B. Neri, and A. Gemignani, ‘‘How breath-control can change your
biomarkers based wearable electrochemical sensors: A mini-review,’’
life: A systematic review on psycho-physiological correlates of slow
Electrochemistry Commun., vol. 140, Jul. 2022, Art. no. 107314, doi:
breathing,’’ Frontiers Hum. Neurosci., vol. 12, p. 353, Sep. 2018, doi:
10.1016/j.elecom.2022.107314.
10.3389/fnhum.2018.00353.
[91] P. C. Ferreira, V. N. Ataíde, C. L. Silva Chagas, L. Angnes,
[71] E. Grossman, A. Grossman, M. Schein, R. Zimlichman, and B. Gavish,
W. K. Tomazelli Coltro, T. R. Longo Cesar Paixão, and W. Reis de Araujo,
‘‘Breathing-control lowers blood pressure,’’ J. Hum. Hypertension,
‘‘Wearable electrochemical sensors for forensic and clinical applica-
vol. 15, no. 4, pp. 263–269, Apr. 2001, doi: 10.1038/sj.jhh.1001147.
tions,’’ TrAC Trends Anal. Chem., vol. 119, Oct. 2019, Art. no. 115622,
[72] K. Sasaki and R. Maruyama, ‘‘Consciously controlled breathing doi: 10.1016/j.trac.2019.115622.
decreases the high-frequency component of heart rate variability by [92] M. Sekar, M. Pandiaraj, S. Bhansali, N. Ponpandian, and C. Viswanathan,
inhibiting cardiac parasympathetic nerve activity,’’ Tohoku J. Experim. ‘‘Carbon fiber based electrochemical sensor for sweat cortisol measure-
Med., vol. 233, no. 3, pp. 155–163, Jul. 2014, doi: 10.1620/tjem.233.155. ment,’’ Sci. Rep., vol. 9, no. 1, p. 403, Jan. 2019, doi: 10.1038/s41598-
[73] M. A. Russo, D. M. Santarelli, and D. O’Rourke, ‘‘The physiological 018-37243-w.
effects of slow breathing in the healthy human,’’ Breathe, vol. 13, no. 4, [93] Y. Shu, T. Su, Q. Lu, Z. Shang, Q. Xu, and X. Hu, ‘‘Highly stretchable
pp. 298–309, Dec. 2017, doi: 10.1183/20734735.009817. wearable electrochemical sensor based on Ni–Co MOF nanosheet-
[74] S. Mathôt, ‘‘Pupillometry: Psychology, physiology, and function,’’ decorated Ag/rGO/PU fiber for continuous sweat glucose detection,’’
J. Cognition, vol. 1, no. 1, pp. 1–18, Feb. 2018, doi: 10.5334/joc.18. Anal. Chem., vol. 93, no. 48, pp. 16222–16230, Dec. 2021, doi:
[75] J. Fong. Eye-Opener: Why Do Pupils Dilate in Response to 10.1021/acs.analchem.1c04106.
Emotional States?. Accessed: Aug. 3, 2022. [Online]. Available: [94] Corticotropin-Releasing Hormone (CRH). Accessed: Aug. 11, 2022.
https://www.scientificamerican.com/article/eye-opener-why-do-pupils- [Online]. Available: https://www.news-medical.net/health/Corticotropin-
dialate/ Releasing-Hormone.aspx
[76] J. He, W. Choi, Y. Yang, J. Lu, X. Wu, and K. Peng, ‘‘Detection of [95] M. Falk, C. Psotta, S. Cirovic, and S. Shleev, ‘‘Non-invasive electrochemi-
driver drowsiness using wearable devices: A feasibility study of the cal biosensors operating in human physiological fluids,’’ Sensors, vol. 20,
proximity sensor,’’ Appl. Ergonom., vol. 65, pp. 473–480, Nov. 2017, doi: no. 21, p. 6352, Nov. 2020, doi: 10.3390/s20216352.
10.1016/j.apergo.2017.02.016. [96] M. Sekar, R. Sriramprabha, P. K. Sekhar, S. Bhansali, N. Ponpan-
[77] C. Jyotsna and J. Amudha, ‘‘Eye gaze as an indicator for stress dian, M. Pandiaraj, and C. Viswanathan, ‘‘Review—Towards wearable
level analysis in students,’’ in Proc. Int. Conf. Adv. Comput., Com- sensor platforms for the electrochemical detection of cortisol,’’ J.
mun. Inform., Sep. 2018, pp. 1588–1593, doi: 10.1109/ICACCI.2018. Electrochemical Soc., vol. 167, no. 6, Mar. 2020, Art. no. 067508, doi:
8554715. 10.1149/1945-7111/ab7e24.

VOLUME 12, 2024 38237


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[97] J. R. Windmiller and J. Wang, ‘‘Wearable electrochemical sensors [117] UCLA. Sweating the Small Stuff: Smartwatch Developed at UCLA Mea-
and biosensors: A review,’’ Electroanalysis, vol. 25, no. 1, pp. 29–46, sures Key Stress Hormone. Accessed: Aug. 17, 2022. [Online]. Available:
Jan. 2013, doi: 10.1002/elan.201200349. https://newsroom.ucla.edu/releases/cortisol-sensing-smartwatch
[98] B. D. Hodge, T. Sanvictores, and R. T. Brodell, ‘‘Anatomy, [118] A. Y. Chang and B. Purt, ‘‘Biochemistry, tear film,’’ in StatPearls. Trea-
skin sweat glands,’’ in StatPearls. Treasure Island, FL, USA: sure Island, FL, USA: StatPearls Publishing, 2022. [Online]. Available:
StatPearls Publishing, 2022. [Online]. Available: http://www.ncbi. http://www.ncbi.nlm.nih.gov/books/NBK572136/
nlm.nih.gov/books/NBK482278/ [119] D. A. Dartt, ‘‘Physiology of tear production,’’ in The Dry Eye: A
[99] L. B. Baker, ‘‘Physiology of sweat gland function: The roles of sweating Comprehensive Guide, M. A. Lemp and R. Marquardt, Eds. Berlin,
and sweat composition in human health,’’ Temperature, vol. 6, no. 3, Heidelberg: Springer, 1992, pp. 65–99, doi: 10.1007/978-3-642-58130-
pp. 211–259, Jul. 2019, doi: 10.1080/23328940.2019.1632145. 4_4.
[100] J. Heikenfeld, ‘‘Non-invasive analyte access and sensing through eccrine [120] S. Gelstein, Y. Yeshurun, L. Rozenkrantz, S. Shushan, I. Frumin, Y. Roth,
sweat: Challenges and outlook circa 2016,’’ Electroanalysis, vol. 28, and N. Sobel, ‘‘Human tears contain a chemosignal,’’ Science, vol. 331,
no. 6, pp. 1242–1249, Jun. 2016, doi: 10.1002/elan.201600018. no. 6014, pp. 226–230, Jan. 2011, doi: 10.1126/science.1198331.
[101] M. J. Nunes, C. M. Cordas, J. J. G. Moura, J. P. Noronha, and
[121] M. Amorim, B. Martins, F. Caramelo, C. Gonçalves, G. Trindade,
L. C. Branco, ‘‘Screening of potential stress biomarkers in sweat
J. Simão, P. Barreto, I. Marques, E. C. Leal, E. Carvalho, F. Reis,
associated with sports training,’’ Sports Med.-Open, vol. 7, no. 1,
T. Ribeiro-Rodrigues, H. Girão, P. Rodrigues-Santos, C. Farinha,
pp. 1–10, Jan. 2021, doi: 10.1186/s40798-020-00294-3.
A. F. Ambrósio, R. Silva, and R. Fernandes, ‘‘Putative biomarkers in
[102] P. J. Derbyshire, H. Barr, F. Davis, and S. P. J. Higson, ‘‘Lactate in human tears for diabetic retinopathy diagnosis,’’ Frontiers Med., vol. 9, pp. 1–22,
sweat: A critical review of research to the present day,’’ J. Physiolog. Sci., May 2022, doi: 10.3389/fmed.2022.873483.
vol. 62, no. 6, pp. 429–440, Nov. 2012, doi: 10.1007/s12576-012-0213-z.
[122] K. Nishtala, N. Pahuja, R. Shetty, R. M. M. A. Nuijts, and A. Ghosh,
[103] J. Moyer, D. Wilson, I. Finkelshtein, B. Wong, and R. Potts, ‘‘Correlation
‘‘Tear biomarkers for keratoconus,’’ Eye Vis., vol. 3, no. 1, pp. 45–49,
between sweat glucose and blood glucose in subjects with diabetes,’’
Aug. 2016, doi: 10.1186/s40662-016-0051-9.
Diabetes Technol. Therapeutics, vol. 14, no. 5, pp. 398–402, May 2012,
doi: 10.1089/dia.2011.0262. [123] A. Chiva, ‘‘Tear biomarkers in dry eye disease,’’ Eur. Ophthalmic Rev.,
[104] E. Russell, G. Koren, M. Rieder, and S. H. M. Van Uum, ‘‘The detection vol. 13, no. 1, p. 21, 2019, doi: 10.17925/eor.2019.13.1.21.
of cortisol in human sweat: Implications for measurement of cortisol in [124] F. Bachhuber, A. Huss, M. Senel, and H. Tumani, ‘‘Diagnostic
hair,’’ Therapeutic Drug Monitor., vol. 36, no. 1, pp. 30–34, Feb. 2014, biomarkers in tear fluid: From sampling to preanalytical processing,’’
doi: 10.1097/ftd.0b013e31829daa0a. Sci. Rep., vol. 11, no. 1, pp. 1–8, May 2021, doi: 10.1038/s41598-021-
[105] P. Pearlmutter, G. DeRose, C. Samson, N. Linehan, Y. Cen, L. Begdache, 89514-8.
D. Won, and A. Koh, ‘‘Sweat and saliva cortisol response to stress and [125] M. Roda, C. Ciavarella, G. Giannaccare, and P. Versura, ‘‘Biomarkers in
nutrition factors,’’ Sci. Rep., vol. 10, no. 1, p. 19050, Nov. 2020, doi: tears and ocular surface: A window for neurodegenerative diseases,’’ Eye
10.1038/s41598-020-75871-3. Contact Lens: Sci. Clin. Pract., vol. 46, no. 2, pp. S129–S134, Mar. 2020,
[106] R. M. Torrente-Rodríguez, J. Tu, Y. Yang, J. Min, M. Wang, Y. Song, doi: 10.1097/icl.0000000000000663.
Y. Yu, C. Xu, C. Ye, W. W. IsHak, and W. Gao, ‘‘Investigation of [126] P. K. Coyle and P. A. Sibony, ‘‘Tear analysis in multiple sclerosis,’’
cortisol dynamics in human sweat using a graphene-based wireless Neurology, vol. 36, no. 4, pp. 1–20, Apr. 1986, doi: 10.1212/wnl.
mHealth system,’’ Matter, vol. 2, no. 4, pp. 921–937, Apr. 2020, doi: 36.4.547.
10.1016/j.matt.2020.01.021. [127] G. Calais, G. Forzy, C. Crinquette, A. Mackowiak, J. de Seze, F. Blanc,
[107] M. Jia, W. M. Chew, Y. Feinstein, P. Skeath, and E. M. Sternberg, C. Lebrun, O. Heinzlef, P. Clavelou, T. Moreau, B. Hennache, H. Zephir,
‘‘Quantification of cortisol in human eccrine sweat by liquid A. Verier, V. Neuville, C. Confavreux, P. Vermersch, and P. Hautecoeur,
chromatography—Tandem mass spectrometry,’’ Analyst, vol. 141, ‘‘Tear analysis in clinically isolated syndrome as new multiple sclerosis
no. 6, pp. 2053–2060, 2016, doi: 10.1039/c5an02387d. criterion,’’ Multiple Sclerosis J., vol. 16, no. 1, pp. 87–92, Jan. 2010, doi:
[108] Verywell Health. New Technology Can Measure Glucose Levels in Sweat, 10.1177/1352458509352195.
No Needles Required. Accessed: Aug. 17, 2022. [Online]. Available: [128] A. Kenny, E. M. Jiménez-Mateos, M. A. Zea-Sevilla, A. Rábano,
https://www.verywellhealth.com/diabetes-glucose-monitor-5208728 P. Gili-Manzanaro, J. H. M. Prehn, D. C. Henshall, J. Ávila, T. Engel,
[109] Phys. World. Silk-Based Sensor Analyses Sweat Biomarkers and F. Hernández, ‘‘Proteins and microRNAs are differentially expressed
in Real Time. Accessed: Aug. 17, 2022. [Online]. Available: in tear fluid from patients with Alzheimer’s disease,’’ Sci. Rep., vol. 9,
https://physicsworld.com/a/silk-based-sensor-analyses-sweat- no. 1, Oct. 2019, Art. no. 1, doi: 10.1038/s41598-019-51837-y.
biomarkers-in-real-time/ [129] M. Boerger, S. Funke, A. Leha, A.-E. Roser, A.-K. Wuestemann,
[110] A. M. V. Mohan, V. Rajendran, R. K. Mishra, and M. Jayaraman, ‘‘Recent F. Maass, M. Bähr, F. Grus, and P. Lingor, ‘‘Proteomic analysis of
advances and perspectives in sweat based wearable electrochemical tear fluid reveals disease-specific patterns in patients with Parkinson’s
sensors,’’ TrAC Trends Anal. Chem., vol. 131, Oct. 2020, Art. no. 116024, disease—A pilot study,’’ Parkinsonism Rel. Disorders, vol. 63, pp. 3–9,
doi: 10.1016/j.trac.2020.116024. Jun. 2019, doi: 10.1016/j.parkreldis.2019.03.001.
[111] A. J. Bandodkar, W. J. Jeang, R. Ghaffari, and J. A. Rogers, ‘‘Wearable [130] C. Tamer, I. M. Melek, T. Duman, and H. Öksüz, ‘‘Tear film tests
sensors for biochemical sweat analysis,’’ Annu. Rev. Anal. Chem., in Parkinson’s disease patients,’’ Ophthalmology, vol. 112, no. 10,
vol. 12, no. 1, pp. 1–22, Jun. 2019, doi: 10.1146/annurev-anchem- pp. 1795.e1–1795.e8, Oct. 2005, doi: 10.1016/j.ophtha.2005.04.025.
061318-114910.
[131] Stress Biomarkers in the Tear Film—Southern Cross
[112] Y. Qiao, L. Qiao, Z. Chen, B. Liu, L. Gao, and L. Zhang, ‘‘Wearable sensor
University. Accessed: Aug. 19, 2022. [Online]. Available: https://
for continuous sweat biomarker monitoring,’’ Chemosensors, vol. 10,
researchportal.scu.edu.au/esploro/outputs/doctoral/Stress-biomarkers-
no. 7, p. 273, Jul. 2022, doi: 10.3390/chemosensors10070273.
in-the-tear-film/991012821564402368
[113] W. Ji, J. Zhu, W. Wu, N. Wang, J. Wang, J. Wu, Q. Wu, X. Wang,
C. Yu, G. Wei, L. Li, and F. Huo, ‘‘Wearable sweat biosensors refresh [132] D. K. Sen and G. S. Sarin, ‘‘Tear glucose levels in normal people and
personalized health/medical diagnostics,’’ Research, vol. 2021, pp. 1–15, in diabetic patients,’’ Brit. J. Ophthalmol., vol. 64, no. 9, pp. 693–695,
Oct. 2021, doi: 10.34133/2021/9757126. Sep. 1980.
[114] Stanford University. (2018). A Wearable Device Measures Cortisol in [133] A. Kijlstra and A. Kuizenga, ‘‘Analysis and function of the human tear
Sweat. Stanford News. Accessed: Aug. 17, 2022. [Online]. Available: proteins,’’ Adv. Exp. Med. Biol., vol. 350, pp. 299–308, Apr. 1994, doi:
https://news.stanford.edu/2018/07/20/wearable-device-measures- 10.1007/978-1-4615-2417-5_51.
cortisol-sweat/ [134] M. Ku, J. Kim, J.-E. Won, W. Kang, Y.-G. Park, J. Park, J.-H. Lee,
[115] J. Xu, Y. Fang, and J. Chen, ‘‘Wearable biosensors for non-invasive J. Cheon, H. H. Lee, and J.-U. Park, ‘‘Smart, soft contact lens for wireless
sweat diagnostics,’’ Biosensors, vol. 11, no. 8, p. 245, Jul. 2021, doi: immunosensing of cortisol,’’ Sci. Adv., vol. 6, no. 28, pp. 1–17, Jul. 2020,
10.3390/bios11080245. doi: 10.1126/sciadv.abb2891.
[116] New Device Can Measure Glucose in Sweat With the Touch [135] Smart Contact Lens Might Measure Cortisol Levels in Cushing’s
of a Fingertip. Accessed: Aug. 17, 2022. [Online]. Available: Patients. Accessed: Aug. 12, 2022. [Online]. Available:
https://www.news-medical.net/news/20210505/New-device-can- https://cushingsdiseasenews.com/news/smart-contact-lens-shows-
measure-glucose-in-sweat-with-the-touch-of-a-fingertip.aspx ability-to-measure-cortisol-levels-in-tears/

38238 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[136] Z. Ballard, S. Bazargan, D. Jung, S. Sathianathan, A. Clemens, [154] S. Perera, M. Uddin, and J. A. Hayes, ‘‘Salivary lysozyme: A noninvasive
D. Shir, S. Al-Hashimi, and A. Ozcan, ‘‘Contact lens-based lysozyme marker for the study of the effects of stress on natural immunity,’’
detection in tear using a mobile sensor,’’ Lab Chip, vol. 20, no. 8, Int. J. Behav. Med., vol. 4, no. 2, pp. 170–178, Jun. 1997, doi:
pp. 1493–1502, 2020, doi: 10.1039/c9lc01039d. 10.1207/s15327558ijbm0402_5.
[137] S. Iguchi, H. Kudo, T. Saito, M. Ogawa, H. Saito, K. Otsuka, [155] E. Aardal-Eriksson, T. E. Eriksson, A.-C. Holm, and T. Lundin, ‘‘Salivary
A. Funakubo, and K. Mitsubayashi, ‘‘A flexible and wearable biosensor cortisol and serum prolactin in relation to stress rating scales in a group of
for tear glucose measurement,’’ Biomed. Microdevices, vol. 9, no. 4, rescue workers,’’ Biol. Psychiatry, vol. 46, no. 6, pp. 850–855, Sep. 1999,
pp. 603–609, Jul. 2007, doi: 10.1007/s10544-007-9073-3. doi: 10.1016/s0006-3223(98)00381-3.
[138] (2020). Smart Contact Lenses Successfully Detect Glucose [156] S. G. Lindell, S. J. Suomi, S. Shoaf, J. D. Higley, and M. Linnoila,
Levels. Accessed: Aug. 19, 2022. [Online]. Available: https://www. ‘‘Salivary prolactin as a marker for central serotonin turnover,’’ Biol. Psy-
diabetes.co.uk/news/2020/may/smart-contact-lenses-successfully- chiatry, vol. 46, no. 4, pp. 568–572, Aug. 1999, doi: 10.1016/s0006-
detect-glucose-levels.html 3223(99)00037-2.
[139] H. Yao, A. J. Shum, M. Cowan, I. Lähdesmäki, and B. A. Parviz, [157] Xtalks. Researchers Develop World’s First Glucose Saliva Test
‘‘A contact lens with embedded sensor for monitoring tear glucose level,’’ for Diabetes. Accessed: Aug. 16, 2022. [Online]. Available:
Biosensors Bioelectron., vol. 26, no. 7, pp. 3290–3296, Mar. 2011, doi: https://xtalks.com/researchers-develop-worlds-first-glucose-saliva-
10.1016/j.bios.2010.12.042. test-for-diabetes-2781/
[140] L. Thau, J. Gandhi, and S. Sharma, ‘‘Physiology, cortisol,’’ in Stat- [158] R. Tiongco, A. Bituin, E. Arceo, N. Rivera, and E. Singian, ‘‘Salivary
Pearls. Treasure Island, FL, USA: StatPearls Publishing, 2022. [Online]. glucose as a non-invasive biomarker of type 2 diabetes mellitus,’’
Available: http://www.ncbi.nlm.nih.gov/books/NBK538239/ J. Clin. Experim. Dentistry, vol. 10, no. 9, p. e902, Sep. 2018, doi:
10.4317/jced.55009.
[141] L. A. Soares Nunes, S. Mussavira, and O. Sukumaran Bindhu, ‘‘Clinical
and diagnostic utility of saliva as a non-invasive diagnostic fluid: A [159] W. Zhang, Y. Du, and M. L. Wang, ‘‘Noninvasive glucose monitoring
systematic review,’’ Biochemia Medica, vol. 25, no. 2, pp. 177–192, using saliva nano-biosensor,’’ Sens. Bio-Sensing Res., vol. 4, pp. 23–29,
Jun. 2015, doi: 10.11613/bm.2015.018. Jun. 2015, doi: 10.1016/j.sbsr.2015.02.002.
[142] R. F. Vining, R. A. McGinley, J. J. Maksvytis, and K. Y. Ho, ‘‘Salivary [160] G. P. Chrousos, ‘‘The concepts of stress and stress system
cortisol: A better measure of adrenal cortical function than serum disorders. Overview of physical and behavioral homeostasis,’’
cortisol,’’ Ann. Clin. Biochemistry: Int. J. Lab. Med., vol. 20, no. 6, J. Amer. Med. Assoc., vol. 267, no. 9, pp. 1244–1252, Mar. 1992.
pp. 329–335, Nov. 1983, doi: 10.1177/000456328302000601. [161] MIT’s Wearable Lab Continuously Monitors Biomarkers
Using Saliva. Accessed: Aug. 19, 2022. [Online]. Available:
[143] Å. M. Hansen, A. H. Garde, and R. Persson, ‘‘Sources of bio-
https://www.hackster.io/news/mit-s-wearable-lab-continuously-
logical and methodological variation in salivary cortisol and their
monitors-biomarkers-using-saliva-87c93aa2d45
impact on measurement among healthy adults: A review,’’ Scan-
din. J. Clin. Lab. Invest., vol. 68, no. 6, pp. 448–458, Jan. 2008, doi: [162] Q. Wang, Q. Yu, Q. Lin, and Y. Duan, ‘‘Emerging salivary biomarkers
10.1080/00365510701819127. by mass spectrometry,’’ Clinica Chim. Acta, vol. 438, pp. 214–221,
Jan. 2015, doi: 10.1016/j.cca.2014.08.037.
[144] G. Briguglio, M. Teodoro, S. Italia, F. Verduci, M. Pollicino, M. Coco,
[163] M. Zea, F. G. Bellagambi, H. Ben Halima, N. Zine, N. Jaffrezic-Renault,
A. De Vita, E. Micali, A. Alibrandi, G. Lembo, C. Costa, and C. Fenga,
R. Villa, G. Gabriel, and A. Errachid, ‘‘Electrochemical sensors for
‘‘Salivary biomarkers and work-related stress in night shift workers,’’
cortisol detections: Almost there,’’ TrAC Trends Anal. Chem., vol. 132,
Int. J. Environ. Res. Public Health, vol. 18, no. 6, p. 3184, Mar. 2021,
Nov. 2020, Art. no. 116058, doi: 10.1016/j.trac.2020.116058.
doi: 10.3390/ijerph18063184.
[164] V. Mani, T. Beduk, W. Khushaim, A. E. Ceylan, S. Timur, O. S. Wolfbeis,
[145] N. Ivković, M. Račić, D. Popović -Grubač, and B. Davidović,
and K. N. Salama, ‘‘Electrochemical sensors targeting salivary biomark-
‘‘Biomarkers of stress in saliva / biomarkeri stresa u pljuvački,’’ Acta
ers: A comprehensive review,’’ TrAC Trends Anal. Chem., vol. 135,
Facultatis Medicae Naissensis, vol. 32, no. 2, pp. 91–99, Jun. 2015, doi:
Feb. 2021, Art. no. 116164, doi: 10.1016/j.trac.2020.116164.
10.1515/afmnai-2015-0010.
[165] Cleveland Clinic. Urinary System. Accessed: Aug. 17, 2022. [Online].
[146] L. Zhang, H. Xiao, and D. T. Wong, ‘‘Salivary biomarkers for clinical
Available: https://my.clevelandclinic.org/health/articles/21197-urinary-
applications,’’ Mol. Diagnosis Therapy, vol. 13, no. 4, pp. 245–259,
system
Aug. 2009, doi: 10.1007/bf03256330.
[166] S. G. Coca and C. R. Parikh, ‘‘Urinary biomarkers for acute kidney injury:
[147] R. T. Chatterton, K. M. Vogelsong, Y. Lu, A. B. Ellman, and Perspectives on translation,’’ Clin. J. Amer. Soc. Nephrology, vol. 3, no. 2,
G. A. Hudgens, ‘‘Salivary α-amylase as a measure of endogenous pp. 481–490, Mar. 2008, doi: 10.2215/cjn.03520807.
adrenergic activity,’’ Clin. Physiol., vol. 16, no. 4, pp. 433–448, Jul. 1996,
[167] J. Jing and Y. Gao, ‘‘Urine biomarkers in the early stages of diseases:
doi: 10.1111/j.1475-097x.1996.tb00731.x.
Current status and perspective,’’ Discov. Med., vol. 25, no. 136, pp. 57–65,
[148] K. T. Kivlighan and D. A. Granger, ‘‘Salivary α-amylase response to Feb. 2018.
competition: Relation to gender, previous experience, and attitudes,’’ [168] H. Su, F. Sun, Z. Lu, J. Zhang, W. Zhang, and J. Liu, ‘‘A wearable
Psychoneuroendocrinology, vol. 31, no. 6, pp. 703–714, Jul. 2006, doi: sensing system based on smartphone and diaper to detect urine in-situ for
10.1016/j.psyneuen.2006.01.007. patients with urinary incontinence,’’ Sens. Actuators B, Chem., vol. 357,
[149] U. Nater, R. Lamarca, L. Florin, A. Moses, W. Langhans, Apr. 2022, Art. no. 131459, doi: 10.1016/j.snb.2022.131459.
M. Koller, and U. Ehlert, ‘‘Stress-induced changes in human salivary [169] J. Zhang, J. Liu, H. Su, F. Sun, Z. Lu, and A. Su, ‘‘A wearable
alpha-amylase activity—Associations with adrenergic activity,’’ self-powered biosensor system integrated with diaper for detecting the
Psychoneuroendocrinology, vol. 31, no. 1, pp. 49–58, Jan. 2006, doi: urine glucose of diabetic patients,’’ Sens. Actuators B, Chem., vol. 341,
10.1016/j.psyneuen.2005.05.010. Aug. 2021, Art. no. 130046, doi: 10.1016/j.snb.2021.130046.
[150] J. Strahler, A. Mueller, F. Rosenloecher, C. Kirschbaum, and N. Rohleder, [170] E. M. Rosen, T. J. van Erve, J. Boss, S. Sathyanarayana, E. S. Barrett,
‘‘Salivary α-amylase stress reactivity across different age groups,’’ R. H. N. Nguyen, N. R. Bush, G. L. Milne, T. F. McElrath, S. H. Swan,
Psychophysiology, vol. 47, no. 3, pp. 587–595, May 2010, doi: and K. K. Ferguson, ‘‘Urinary oxidative stress biomarkers and accelerated
10.1111/j.1469-8986.2009.00957.x. time to spontaneous delivery,’’ Free Radical Biol. Med., vol. 130,
[151] S. Cozma, L. C. Dima-Cozma, C. M. Ghiciuc, V. Pasquali, A. Saponaro, pp. 419–425, Jan. 2019, doi: 10.1016/j.freeradbiomed.2018.11.011.
and F. R. Patacchioli, ‘‘Salivary cortisol and α-amylase: Subclinical [171] M. Martínez-Estrada, R. Fernández-García, and I. Gil, ‘‘A wearable
indicators of stress as cardiometabolic risk,’’ Brazilian J. Med. Biol. Res., system to detect urine leakage based on a textile sensor,’’ in Proc. IEEE
vol. 50, no. 2, pp. 1–18, Feb. 2017, doi: 10.1590/1414-431x20165577. Int. Conf. Flexible Printable Sensors Syst. (FLEPS), Aug. 2020, pp. 1–3,
[152] M. H. Laudat, S. Cerdas, C. Fournier, D. Guiban, B. Guilhaume, doi: 10.1109/FLEPS49123.2020.9239554.
and J. P. Luton, ‘‘Salivary cortisol measurement: A practical approach [172] A. A. H. Gadalla, I. M. Friberg, A. Kift-Morgan, J. Zhang, M. Eberl,
to assess pituitary-adrenal function,’’ J. Clin. Endocrinol. Metabolism, N. Topley, I. Weeks, S. Cuff, M. Wootton, M. Gal, G. Parekh, P. Davis,
vol. 66, no. 2, pp. 343–348, Feb. 1988, doi: 10.1210/jcem-66-2-343. C. Gregory, K. Hood, K. Hughes, C. Butler, and N. A. Francis,
[153] D. Bozovic, M. Racic, and N. Ivkovic, ‘‘Salivary cortisol levels as ‘‘Identification of clinical and urine biomarkers for uncomplicated
a biological marker of stress reaction,’’ Med. Arch., vol. 67, p. 374, urinary tract infection using machine learning algorithms,’’ Sci. Rep.,
2013. vol. 9, no. 1, pp. 1–10, Dec. 2019, doi: 10.1038/s41598-019-55523-x.

VOLUME 12, 2024 38239


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[173] L. Johnston, G. Wang, K. Hu, C. Qian, and G. Liu, ‘‘Advances [190] R. Zhao, C. Wang, F. Lu, L. Du, Z. Fang, X. Guo, J.-T. Liu,
in biosensors for continuous glucose monitoring towards wearables,’’ C.-J. Chen, and Z. Zhao, ‘‘A flexible interdigital electrode used in skin
Frontiers Bioeng. Biotechnol., vol. 9, pp. 1–12, Aug. 2021. penetration promotion and evaluation with electroporation and reverse
[174] C. Kolluru, M. Williams, J. Chae, and M. R. Prausnitz, ‘‘Recruitment iontophoresis synergistically,’’ Sensors, vol. 18, no. 5, p. 1431, May 2018,
and collection of dermal interstitial fluid using a microneedle patch,’’ doi: 10.3390/s18051431.
Adv. Healthcare Mater., vol. 8, no. 3, Feb. 2019, Art. no. e1801262, doi: [191] E. De la Paz, A. Barfidokht, S. Rios, C. Brown, E. Chao, and
10.1002/adhm.201801262. J. Wang, ‘‘Extended noninvasive glucose monitoring in the interstitial
[175] Definition of Interstitial Fluid—NCI Dictionary of Cancer fluid using an epidermal biosensing patch,’’ Anal. Chem., vol. 93, no. 37,
Terms—NCI. Accessed: Aug. 18, 2022. [Online]. Available: pp. 12767–12775, Sep. 2021, doi: 10.1021/acs.analchem.1c02887.
https://www.cancer.gov/publications/dictionaries/cancer- [192] A. J. Bandodkar, W. Jia, C. Yardımcı, X. Wang, J. Ramirez, and
terms/def/interstitial-fluid J. Wang, ‘‘Tattoo-based noninvasive glucose monitoring: A proof-of-
[176] M. M. Niedzwiecki, P. Samant, D. I. Walker, V. Tran, D. P. Jones, concept study,’’ Anal. Chem., vol. 87, no. 1, pp. 394–398, Jan. 2015, doi:
M. R. Prausnitz, and G. W. Miller, ‘‘Human suction blister fluid 10.1021/ac504300n.
composition determined using high-resolution metabolomics,’’ [193] H. C. Ates, P. Q. Nguyen, L. Gonzalez-Macia, E. Morales-Narváez,
Anal. Chem., vol. 90, no. 6, pp. 3786–3792, Mar. 2018, doi: F. Güder, J. J. Collins, and C. Dincer, ‘‘End-to-end design of wearable
10.1021/acs.analchem.7b04073. sensors,’’ Nature Rev. Mater., vol. 7, no. 11, pp. 887–907, Jul. 2022, doi:
[177] G. Baronzio, G. Parmar, M. Baronzio, and M. Kiselevsky, ‘‘Tumor 10.1038/s41578-022-00460-x.
interstitial fluid: Proteomic determination as a possible source of [194] M. Venugopal, S. K. Arya, G. Chornokur, and S. Bhansali, ‘‘A realtime
biomarkers,’’ Cancer Genomics Proteomics, vol. 11, no. 5, pp. 55–64, and continuous assessment of cortisol in ISF using electrochemical
Sep. 2014. impedance spectroscopy,’’ Sens. Actuators A, Phys., vol. 172, no. 1,
[178] P. Gromov, I. Gromova, C. J. Olsen, V. Timmermans-Wielenga, pp. 154–160, Dec. 2011, doi: 10.1016/j.sna.2011.04.028.
M.-L. Talman, R. R. Serizawa, and J. M. A. Moreira, ‘‘Tumor interstitial [195] J. Kim, J. R. Sempionatto, S. Imani, M. C. Hartel, A. Barfidokht,
fluid—A treasure trove of cancer biomarkers,’’ Biochimica et Biophysica G. Tang, A. S. Campbell, P. P. Mercier, and J. Wang, ‘‘Simultaneous
Acta (BBA)—Proteins Proteomics, vol. 1834, no. 11, pp. 2259–2270, monitoring of sweat and interstitial fluid using a single wearable
Nov. 2013, doi: 10.1016/j.bbapap.2013.01.013. biosensor platform,’’ Adv. Sci., vol. 5, no. 10, Oct. 2018, Art. no. 1800880,
[179] T. Kiang, S. Ranamukhaarachchi, and M. Ensom, ‘‘Revolutionizing doi: 10.1002/advs.201800880.
therapeutic drug monitoring with the use of interstitial fluid and [196] (2014). Estimation and Statistical Analysis of Human Voice
microneedles technology,’’ Pharmaceutics, vol. 9, no. 4, p. 43, Oct. 2017, Parameters to Investigate the Influence of Psychological Stress
doi: 10.3390/pharmaceutics9040043. and to Determine the Vocal Tract Transfer Function of an
[180] A Device for Opioid Biomarker Sensing From Interstitial Individual. Accessed: Aug. 25, 2022. [Online]. Available: https://www.
Fluid. Accessed: Aug. 20, 2022. [Online]. Available: hindawi.com/journals/jcnc/2014/290147/
https://www.fda.gov/science-research/fda-science-forum/device-opioid- [197] V. Tumanova and N. Backes, ‘‘Autonomic nervous system response
biomarker-sensing-interstitial-fluid to speech production in stuttering and normally fluent preschool-age
[181] C. Kolluru, M. Williams, J. S. Yeh, R. K. Noel, J. Knaack, and children,’’ J. Speech, Lang., Hearing Res., vol. 62, no. 11, pp. 4030–4044,
M. R. Prausnitz, ‘‘Monitoring drug pharmacokinetics and immunologic Nov. 2019, doi: 10.1044/2019_jslhr-s-19-0121.
biomarkers in dermal interstitial fluid using a microneedle patch,’’ [198] Frontiers | Autonomic Nervous System Activity During a
Biomed. Microdevices, vol. 21, no. 1, pp. 1–5, Feb. 2019, doi: Speech Task. Accessed: Aug. 25, 2022. [Online]. Available:
10.1007/s10544-019-0363-3. https://www.frontiersin.org/articles/10.3389/fnins.2019.00406/full
[182] M. Parrilla, U. Detamornrat, J. Domínguez-Robles, R. F. Donnelly, [199] Full Article: Stress Measurement Using Speech: Recent Advancements,
and K. De Wael, ‘‘Wearable hollow microneedle sensing patches for Validation Issues, and Ethical and Privacy Considerations. Accessed:
the transdermal electrochemical monitoring of glucose,’’ Talanta, Aug. 25, 2022. [Online]. Available: https://www.
vol. 249, Nov. 2022, Art. no. 123695, doi: 10.1016/j.talanta.2022. tandfonline.com/doi/full/10.1080/10253890.2019.1584180
123695. [200] M. Kappen, K. Hoorelbeke, N. Madhu, K. Demuynck, and M.-
[183] J. Kim, A. S. Campbell, and J. Wang, ‘‘Wearable non-invasive epidermal A. Vanderhasselt, ‘‘Speech as an indicator for psychosocial stress:
glucose sensors: A review,’’ Talanta, vol. 177, pp. 163–170, Jan. 2018, A network analytic approach,’’ Behav. Res. Methods, vol. 54, no. 2,
doi: 10.1016/j.talanta.2017.08.077. pp. 910–921, Apr. 2022, doi: 10.3758/s13428-021-01670-x.
[184] X. Jin, G. Li, T. Xu, L. Su, D. Yan, and X. Zhang, ‘‘Fully inte- [201] J. H. L. Hansen and S. Patil, ‘‘Speech under stress: Analysis,
grated flexible biosensor for wearable continuous glucose monitoring,’’ modeling and recognition,’’ in Speaker Classification I: Fundamentals,
Biosensors Bioelectron., vol. 196, Jan. 2022, Art. no. 113760, doi: Features, and Methods (Lecture Notes in Computer Science). Heidelberg,
10.1016/j.bios.2021.113760. Germany: Springer, 2007, pp. 108–137, doi: 10.1007/978-3-540-74200-
[185] R. He, Y. Niu, Z. Li, A. Li, H. Yang, F. Xu, and F. Li, ‘‘A hydrogel 5_6.
microneedle patch for point-of-care testing based on skin interstitial [202] G. M. Slavich, S. Taylor, and R. W. Picard, ‘‘Stress measurement using
fluid,’’ Adv. Healthcare Mater., vol. 9, no. 4, Feb. 2020, Art. no. 4, doi: speech: Recent advancements, validation issues, and ethical and privacy
10.1002/adhm.201901201. considerations,’’ Stress, vol. 22, no. 4, pp. 408–413, Jul. 2019, doi:
[186] P. P. Samant and M. R. Prausnitz, ‘‘Mechanisms of sampling 10.1080/10253890.2019.1584180.
interstitial fluid from skin using a microneedle patch,’’ [203] B. Shahidi, A. Haight, and K. Maluf, ‘‘Differential effects of mental
Proc. Nat. Acad. Sci. USA, vol. 115, no. 18, pp. 4583–4588, May 2018, concentration and acute psychosocial stress on cervical muscle activity
doi: 10.1073/pnas.1716772115. and posture,’’ J. Electromyogr. Kinesiol., vol. 23, no. 5, pp. 1082–1089,
[187] H. Chang, M. Zheng, X. Yu, A. Than, R. Z. Seeni, R. Kang, J. Tian, Oct. 2013, doi: 10.1016/j.jelekin.2013.05.009.
D. P. Khanh, L. Liu, P. Chen, and C. Xu, ‘‘A swellable microneedle [204] M. Landers, G. Barker, S. Wallentine, J. W. McWhorter, and C. Peel,
patch to rapidly extract skin interstitial fluid for timely metabolic ‘‘A comparison of tidal volume, breathing frequency, and minute
analysis,’’ Adv. Mater., vol. 29, no. 37, Oct. 2017, Art. no. 1702243, doi: ventilation between two sitting postures in healthy adults,’’ Physio-
10.1002/adma.201702243. therapy Theory Pract., vol. 19, no. 2, pp. 109–119, Jul. 2009, doi:
[188] J. Chen, M. Wang, Y. Ye, Z. Yang, Z. Ruan, and N. Jin, ‘‘Fabrication of 10.1080/09593980307958.
sponge-forming microneedle patch for rapidly sampling interstitial fluid [205] S. Nair, M. Sagar, J. Sollers, N. Consedine, and E. Broadbent, ‘‘Do
for analysis,’’ Biomed. Microdevices, vol. 21, no. 3, pp. 1–10, Jul. 2019, slumped and upright postures affect stress responses? A randomized
doi: 10.1007/s10544-019-0413-x. trial,’’ Health Psychol., vol. 34, no. 6, pp. 632–641, Jun. 2015, doi:
[189] N. Kashaninejad, A. Munaz, H. Moghadas, S. Yadav, M. Umer, and 10.1037/hea0000146.
N.-T. Nguyen, ‘‘Microneedle arrays for sampling and sensing skin [206] M. Ardito, F. Mascolo, M. Valentini, and F. Dell’Olio, ‘‘Low-cost wireless
interstitial fluid,’’ Chemosensors, vol. 9, no. 4, p. 83, Apr. 2021, doi: wearable system for posture monitoring,’’ Electronics, vol. 10, no. 21,
10.3390/chemosensors9040083. p. 2569, Oct. 2021, doi: 10.3390/electronics10212569.

38240 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[207] R. Bootsman, P. Markopoulos, Q. Qi, Q. Wang, and A. A. Timmermans, [227] Shaking Hands (Hand Tremors): 14 Causes and
‘‘Wearable technology for posture monitoring at the workplace,’’ Treatments. Accessed: Aug. 25, 2022. [Online]. Available: https://www.
Int. J. Hum.-Comput. Stud., vol. 132, pp. 99–111, Dec. 2019, doi: medicalnewstoday.com/articles/322195
10.1016/j.ijhcs.2019.08.003. [228] What Causes Shaky Hands and How is it Treated?
[208] Smart Clothing Can Monitor UV Levels, Improve Posture. Accessed: Accessed: Aug. 25, 2022. [Online]. Available:
Aug. 25, 2022. [Online]. Available: https://www.thomasnet. https://www.webmd.com/brain/shaky-hands
com/insights/smart-clothing-can-monitor-uv-levels-improve-posture/ [229] H. Abbas, K. Zahed, L. Alic, Y. Zhu, F. Sasangohar, R. Mehta, M. Lawley,
[209] W. Y. Wong and M. S. Wong, ‘‘Smart garment for trunk posture Q. H. Abbasi, and K. A. Qaraqe, ‘‘A wearable, low-cost hand tremor
monitoring: A preliminary study,’’ Scoliosis, vol. 3, no. 1, pp. 1–9, sensor for detecting hypoglycemic events in diabetic patients,’’ in
May 2008, doi: 10.1186/1748-7161-3-7. Proc. IEEE Int. RF Microw. Conf. (RFM), Dec. 2018, pp. 182–184, doi:
[210] Q. Wang, W. Chen, A. A. A. Timmermans, C. Karachristos, J. B. Martens, 10.1109/RFM.2018.8846546.
and P. Markopoulos, ‘‘Smart rehabilitation garment for posture monitor- [230] H. Jeon, W. Lee, H. Park, H. Lee, S. Kim, H. Kim, B. Jeon, and K. Park,
ing,’’ in Proc. 37th Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. (EMBC), ‘‘Automatic classification of tremor severity in Parkinson’s disease using
Aug. 2015, pp. 5736–5739, doi: 10.1109/EMBC.2015.7319695. a wearable device,’’ Sensors, vol. 17, no. 9, p. 2067, Sep. 2017, doi:
[211] Q. Wang, M. Toeters, W. Chen, A. Timmermans, and P. Markopoulos, 10.3390/s17092067.
‘‘Zishi: A smart garment for posture monitoring,’’ in Proc. CHI [231] M. Delrobaei, S. Memar, M. Pieterman, T. W. Stratton, K. McIsaac, and
Conf. Extended Abstr. Hum. Factors Comput. Syst., May 2016, M. Jog, ‘‘Towards remote monitoring of Parkinson’s disease tremor using
pp. 3792–3795, doi: 10.1145/2851581.2890262. wearable motion capture systems,’’ J. Neurolog. Sci., vol. 384, pp. 38–45,
[212] Y. Jiang, J. An, F. Liang, G. Zuo, J. Yi, C. Ning, H. Zhang, K. Dong, and Jan. 2018, doi: 10.1016/j.jns.2017.11.004.
Z. L. Wang, ‘‘Knitted self-powered sensing textiles for machine learning- [232] B. Vescio, A. Quattrone, R. Nisticò, M. Crasà, and A. Quattrone,
assisted sitting posture monitoring and correction,’’ Nano Res., vol. 15, ‘‘Wearable devices for assessment of tremor,’’ Frontiers Neurol., vol. 12,
no. 9, pp. 8389–8397, Sep. 2022, doi: 10.1007/s12274-022-4409-0. Jun. 2021, Art. no. 680011.
[213] W.-Y. Lin, M.-Y. Lee, and W.-C. Chou, ‘‘The design and devel- [233] R. Stojanovic and D. Karadaglic, ‘‘A LED–LED-based photoplethys-
opment of a wearable posture monitoring vest,’’ in Proc. IEEE mography sensor,’’ Physiological Meas., vol. 28, no. 6, pp. N19–N27,
Int. Conf. Consum. Electron. (ICCE), Jan. 2014, pp. 329–330, doi: Jun. 2007, doi: 10.1088/0967-3334/28/6/n01.
10.1109/ICCE.2014.6776027. [234] M. Ghamari, ‘‘A review on wearable photoplethysmography sensors and
[214] W.-Y. Lin, W.-C. Chou, T.-H. Tsai, C.-C. Lin, and M.-Y. Lee, ‘‘Devel- their potential future applications in health care,’’ Int. J. Biosensors
opment of a wearable instrumented vest for posture monitoring and Bioelectron., vol. 4, no. 4, pp. 195–202, 2018, doi: 10.15406/ijb-
system usability verification based on the technology acceptance model,’’ sbe.2018.04.00125.
Sensors, vol. 16, no. 12, p. 2172, Dec. 2016, doi: 10.3390/s16122172. [235] Photoplethysmography: Technology, Signal Analysis and Applications,
[215] H. Yoon, S. Hwang, D. Jung, S. Choi, K. Joo, J. Choi, Y. Lee, Elsevier, Amsterdam, The Netherlands, 2021.
D.-U. Jeong, and K. Park, ‘‘Estimation of sleep posture using a patch- [236] E. Mejia-Mejia, J. Allen, K. Budidha, C. El-Hajj, P. A. Kyria-
type accelerometer based device,’’ in Proc. 37th Annu. Int. Conf. IEEE cou, and P. H. Charlton, ‘‘Photoplethysmography signal processing
Eng. Med. Biol. Soc. (EMBC), Aug. 2015, pp. 4942–4945, doi: and synthesis,’’ in Photoplethysmography, P. Kyriacou and J. Allen,
10.1109/EMBC.2015.7319500. Eds. Amsterdam, The Netherlands: Elsevier, 2021, pp. 69–146, doi:
[216] P.-Y. Jeng, L.-C. Wang, C.-J. Hu, and D. Wu, ‘‘A wrist sensor sleep 10.1016/B978-0-12-823374-0.00015-3.
posture monitoring system: An automatic labeling approach,’’ Sensors, [237] J. M. M. Torres, A. Ghosh, E. A. Stepanov, and G. Riccardi, ‘‘Heal-
vol. 21, no. 1, p. 258, Jan. 2021, doi: 10.3390/s21010258. T: An efficient PPG-based heart-rate and IBI estimation method during
[217] F. Foerster, M. Smeja, and J. Fahrenberg, ‘‘Detection of posture and physical exercise,’’ in Proc. 24th Eur. Signal Process. Conf. (EUSIPCO),
motion by accelerometry: A validation study in ambulatory monitoring,’’ Aug. 2016, pp. 1438–1442, doi: 10.1109/EUSIPCO.2016.7760486.
Comput. Hum. Behav., vol. 15, no. 5, pp. 571–583, Sep. 1999, doi:
[238] S. Elzeiny and M. Qaraqe, ‘‘Automatic and intelligent stressor identifi-
10.1016/s0747-5632(99)00037-0.
cation based on photoplethysmography analysis,’’ IEEE Access, vol. 9,
[218] D. Curone, G. M. Bertolotti, A. Cristiani, E. L. Secco, and G. Magenes,
pp. 68498–68510, 2021, doi: 10.1109/ACCESS.2021.3077358.
‘‘A real-time and self-calibrating algorithm based on triaxial accelerom-
[239] T. Iqbal, A. Elahi, S. Ganly, W. Wijns, and A. Shahzad,
eter signals for the detection of human posture and activity,’’ IEEE
‘‘Photoplethysmography-based respiratory rate estimation algorithm
Trans. Inf. Technol. Biomed., vol. 14, no. 4, pp. 1098–1105, Jul. 2010,
for health monitoring applications,’’ J. Med. Biol. Eng., vol. 42, no. 2,
doi: 10.1109/TITB.2010.2050696.
pp. 242–252, Apr. 2022, doi: 10.1007/s40846-022-00700-z.
[219] L. Q. Vu, K. H. Kim, L. J. H. Schulze, and S. L. Rajulu, ‘‘Lumbar posture
assessment with fabric strain sensors,’’ Comput. Biol. Med., vol. 118, [240] S. Blok, M. A. Piek, I. I. Tulevski, G. A. Somsen, and M. M. Winter, ‘‘The
Mar. 2020, Art. no. 103624, doi: 10.1016/j.compbiomed.2020.103624. accuracy of heartbeat detection using photoplethysmography technology
in cardiac patients,’’ J. Electrocardiol., vol. 67, pp. 148–157, Jul. 2021,
[220] Y.-R. Guo, X.-C. Zhang, and N. An, ‘‘Monitoring neck posture with flex
doi: 10.1016/j.jelectrocard.2021.06.009.
sensors,’’ in Proc. 9th Int. Conf. Inf. Sci. Technol. (ICIST), Aug. 2019,
pp. 459–463, doi: 10.1109/ICIST.2019.8836806. [241] R. Lei, B. W.-K. Ling, P. Feng, and J. Chen, ‘‘Estimation of heart rate
[221] Y. Cha, H. Kim, and D. Kim, ‘‘Flexible piezoelectric sensor-based and respiratory rate from PPG signal using complementary ensemble
gait recognition,’’ Sensors, vol. 18, no. 2, p. 468, Feb. 2018, doi: empirical mode decomposition with both independent component
10.3390/s18020468. analysis and non-negative matrix factorization,’’ Sensors, vol. 20, no. 11,
[222] J. A. Bell and M. Stigant, ‘‘Development of a fibre optic goniometer p. 3238, Jun. 2020, doi: 10.3390/s20113238.
system to measure lumbar and hip movement to detect activities and their [242] J. D. Campbell, C. G. Pretty, J. G. Chase, and P. J. Bones, ‘‘Near-real-time
lumbar postures,’’ J. Med. Eng. Technol., vol. 31, no. 5, pp. 361–366, detection of pulse oximeter PPG peaks using wavelet decomposition,’’
Jan. 2007, doi: 10.1080/03091900600996735. IFAC-PapersOnLine, vol. 51, no. 27, pp. 146–151, Jan. 2018, doi:
[223] P. Ribeiro, A. R. Soares, R. Girão, M. Neto, and S. Cardoso, ‘‘Spine 10.1016/j.ifacol.2018.11.652.
cop: Posture correction monitor and assistant,’’ Sensors, vol. 20, no. 18, [243] A. Sharma, R. S. Tanwar, Y. Singh, and A. Sharma, ‘‘Heart rate and
p. 5376, Sep. 2020, doi: 10.3390/s20185376. blood pressure measurement based on photoplethysmogram signal using
[224] A. Bhattacharya, J. Warren, J. Teuschler, M. Dimov, M. Medvedovic, fast Fourier transform,’’ Comput. Electr. Eng., vol. 101, Jul. 2022,
and G. Lemasters, ‘‘Development and evaluation of a microprocessor- Art. no. 108057, doi: 10.1016/j.compeleceng.2022.108057.
based ergonomic dosimeter for evaluating carpentry tasks,’’ Appl. [244] R. Lazazzera, Y. Belhaj, and G. Carrault, ‘‘A new wearable device for
Ergonom., vol. 30, no. 6, pp. 543–553, Dec. 1999, doi: 10.1016/s0003- blood pressure estimation using photoplethysmogram,’’ Sensors, vol. 19,
6870(99)00006-x. no. 11, p. 2557, Jun. 2019, doi: 10.3390/s19112557.
[225] W. Y. Wong and M. S. Wong, ‘‘Trunk posture monitoring with inertial [245] C. Hale. FDA Clears ‘Cuffless’ Blood Pressure Tracking Smart-
sensors,’’ Eur. Spine J., vol. 17, no. 5, pp. 743–753, May 2008, doi: watch From Biobeat. Accessed: Nov. 13, 2022. [Online]. Avail-
10.1007/s00586-008-0586-0. able: https://www.fiercebiotech.com/medtech/fda-clears-cuffless-blood-
[226] Cleveland Clinic. Essential Tremor: What it is, Causes, Symp- pressure-tracking-smartwatch-from-biobeat
toms & Treatment. Accessed: Aug. 25, 2022. [Online]. Available: [246] D. H. Sliney, ‘‘What is light? The visible spectrum and beyond,’’ Eye,
https://my.clevelandclinic.org/health/diseases/11886-essential-tremor vol. 30, no. 2, pp. 222–229, Feb. 2016, doi: 10.1038/eye.2015.252.

VOLUME 12, 2024 38241


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[247] (2007). ISO 20473:2007. Accessed: Mar. 12, 2023. [Online]. Available: [266] D. Shastri, M. Papadakis, P. Tsiamyrtzis, B. Bass, and I. Pavlidis,
https://www.iso.org/standard/39482.html ‘‘Perinasal imaging of physiological stress and its affective potential,’’
[248] J. E. Sinex, ‘‘Pulse oximetry: Principles and limitations,’’ Amer. J. Emer- IEEE Trans. Affect. Comput., vol. 3, no. 3, pp. 366–378, Jul. 2012, doi:
gency Med., vol. 17, no. 1, pp. 59–66, Jan. 1999, doi: 10.1016/s0735- 10.1109/T-AFFC.2012.13.
6757(99)90019-0. [267] T. Chen, P. Yuen, M. Richardson, G. Liu, and Z. She, ‘‘Detection of
[249] A. Jubran, ‘‘Pulse oximetry,’’ Crit. Care, vol. 19, no. 1, p. 272, Dec. 2015, psychological stress using a hyperspectral imaging technique,’’ IEEE
doi: 10.1186/s13054-015-0984-8. Trans. Affect. Comput., vol. 5, no. 4, pp. 391–405, Oct. 2014, doi:
[250] M. Nitzan, A. Romem, and R. Koppel, ‘‘Pulse oximetry: Fundamentals 10.1109/TAFFC.2014.2362513.
and technology update,’’ Med. Devices: Evidence Res., p. 231, Jul. 2014, [268] I. Pavlidis, J. Levine, and P. Baukol, ‘‘Thermal imaging for anxiety detec-
doi: 10.2147/mder.s47319. tion,’’ in Proc. IEEE Workshop Comput. Vis. Beyond Visible Spectr., Meth-
[251] T. Liang, Y. Guan, C. Lei, X. Wu, Y. Bai, J. Xiong, and L. Qi, ‘‘Design and ods Appl., Jun. 2000, pp. 104–109, doi: 10.1109/CVBVS.2000.855255.
fabrication of a low-cost thermopile infrared detector,’’ Micromachines, [269] Electroencephalogram (EEG). Accessed: Sep. 20, 2022. [Online].
vol. 12, no. 9, p. 1134, Sep. 2021, doi: 10.3390/mi12091134. Available: https://www.hopkinsmedicine.org/health/treatment-tests-and-
[252] D. D. Soerensen and L. J. Pedersen, ‘‘Infrared skin temperature therapies/electroencephalogram-eeg
measurements for monitoring health in pigs: A review,’’ Acta Veterinaria [270] J. S. Kumar and P. Bhuvaneswari, ‘‘Analysis of electroencephalography
Scandinavica, vol. 57, no. 1, p. 5, Feb. 2015, doi: 10.1186/s13028-015- (EEG) signals and its categorization—A study,’’ Proc. Eng., vol. 38,
0094-2. pp. 2525–2536, Jan. 2012, doi: 10.1016/j.proeng.2012.06.298.
[253] T. Jue and K. Masuda, Application of Near Infrared Spectroscopy in [271] R. Sharma and K. Chopra, ‘‘EEG signal analysis and detection of stress
Biomedicine. Cham, Switzerland: Springer, 2013. using classification techniques,’’ J. Inf. Optim. Sci., vol. 41, no. 1,
[254] D. K. Nguyen, J. Tremblay, P. Pouliot, P. Vannasing, O. Florea, pp. 229–238, Jan. 2020, doi: 10.1080/02522667.2020.1714187.
L. Carmant, F. Lepore, M. Sawan, F. Lesage, and M. Lassonde, [272] M. S. Kalas and B. F. Momin, ‘‘Stress detection and reduction using EEG
‘‘Non-invasive continuous EEG-fNIRS recording of temporal lobe signals,’’ in Proc. Int. Conf. Electr., Electron., Optim. Techn. (ICEEOT),
seizures,’’ Epilepsy Res., vol. 99, nos. 1–2, pp. 112–126, Mar. 2012, doi: Mar. 2016, pp. 471–475, doi: 10.1109/ICEEOT.2016.7755604.
10.1016/j.eplepsyres.2011.10.035. [273] G. Jun and K. G. Smitha, ‘‘EEG based stress level identification,’’
[255] N. K. Schaal, P. Hepp, A. Schweda, O. T. Wolf, and C. Krampe, ‘‘A func- in Proc. IEEE Int. Conf. Syst., Man, Cybern. (SMC), Oct. 2016,
tional near-infrared spectroscopy study on the cortical haemodynamic pp. 3270–3274, doi: 10.1109/SMC.2016.7844738.
responses during the Maastricht acute stress test,’’ Sci. Rep., vol. 9, no. 1, [274] C.-Y. Liao, R.-C. Chen, and S.-K. Tai, ‘‘Emotion stress detection
pp. 1–10, Sep. 2019, doi: 10.1038/s41598-019-49826-2. using EEG signal and deep learning technologies,’’ in Proc. IEEE
[256] P. Pinti, I. Tachtsidis, A. Hamilton, J. Hirsch, C. Aichelburg, Int. Conf. Appl. Syst. Invention (ICASI), Apr. 2018, pp. 90–93, doi:
S. Gilbert, and P. W. Burgess, ‘‘The present and future use of func- 10.1109/ICASI.2018.8394414.
tional near-infrared spectroscopy (fNIRS) for cognitive neuroscience,’’ [275] R. Katmah, F. Al-Shargie, U. Tariq, F. Babiloni, F. Al-Mughairbi,
Ann. New York Acad. Sci., vol. 1464, no. 1, pp. 5–29, Mar. 2020, doi: and H. Al-Nashash, ‘‘A review on mental stress assessment methods
10.1111/nyas.13948. using EEG signals,’’ Sensors, vol. 21, no. 15, p. 5043, Jul. 2021, doi:
[257] A. R. Anwar, M. Muthalib, S. Perrey, A. Galka, O. Granert, S. Wolff, 10.3390/s21155043.
U. Heute, G. Deuschl, J. Raethjen, and M. Muthuraman, ‘‘Effective [276] M. Haak, S. Bos, S. Panic, and L. J. M. Rothkrantz, ‘‘Detecting stress
connectivity of cortical sensorimotor networks during finger movement using eye blinks and brain activity from EEG signals,’’ in Proc. 1st Driver
tasks: A simultaneous fNIRS, fMRI, EEG study,’’ Brain Topography, Car Interact. Interface, 2008, pp. 35–60.
vol. 29, no. 5, pp. 645–660, Sep. 2016, doi: 10.1007/s10548-016-0507- [277] A. J. Casson, D. C. Yates, S. J. M. Smith, J. S. Duncan, and
1. E. Rodriguez-Villegas, ‘‘Wearable electroencephalography,’’ IEEE
[258] S. Knopp, P. Bones, S. Weddell, C. Innes, and R. Jones, ‘‘A wearable Eng. Med. Biol. Mag., vol. 29, no. 3, pp. 44–56, May 2010, doi:
device for measuring eye dynamics in real-world conditions,’’ in 10.1109/MEMB.2010.936545.
Proc. 35th Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. (EMBC), Jul. 2013, [278] C.-T. Lin, C.-H. Chuang, C.-S. Huang, S.-F. Tsai, S.-W. Lu, Y.-H. Chen,
pp. 6615–6618, doi: 10.1109/EMBC.2013.6611072. and L.-W. Ko, ‘‘Wireless and wearable EEG system for evaluating driver
[259] A. Bulling, D. Roggen, and G. Tröster, ‘‘Wearable EOG goggles: Eye- vigilance,’’ IEEE Trans. Biomed. Circuits Syst., vol. 8, no. 2, pp. 165–176,
based interaction in everyday environments,’’ in Proc. CHI Extended Apr. 2014, doi: 10.1109/TBCAS.2014.2316224.
Abstr. Hum. Factors Comput. Syst., Apr. 2009, pp. 3259–3264, doi: [279] G. Li, B.-L. Lee, and W.-Y. Chung, ‘‘Smartwatch-based wearable EEG
10.1145/1520340.1520468. system for driver drowsiness detection,’’ IEEE Sensors J., vol. 15, no. 12,
[260] M. M. Kalas and B. Momin, ‘‘Blink detection using electrooculography pp. 7169–7180, Dec. 2015, doi: 10.1109/JSEN.2015.2473679.
for stress detection,’’ in Proc. Nat. Res. Symp. Comput., vol. 10, pp. 1–20, [280] S. Hwang, H. Jebelli, B. Choi, M. Choi, and S. Lee, ‘‘Measuring
Dec. 2016. workers’ emotional state during construction tasks using wearable
[261] P. A. D. L. Medeiros, G. V. S. D. Silva, F. R. D. S. Fer- EEG,’’ J. Construction Eng. Manage., vol. 144, no. 7, Jul. 2018,
nandes, I. Sánchez-Gendriz, H. W. C. Lins, D. M. D. S. Barros, Art. no. 04018050, doi: 10.1061/(asce)co.1943-7862.0001506.
D. A. P. Nagem, and R. A. D. M. Valentim, ‘‘Efficient machine [281] U. Ha, Y. Lee, H. Kim, T. Roh, J. Bae, C. Kim, and H.-J. Yoo, ‘‘A
learning approach for volunteer eye-blink detection in real-time using wearable EEG-HEG-HRV multimodal system with simultaneous
webcam,’’ Expert Syst. Appl., vol. 188, Feb. 2022, Art. no. 116073, doi: monitoring of tES for mental health management,’’ IEEE
10.1016/j.eswa.2021.116073. Trans. Biomed. Circuits Syst., vol. 9, no. 6, pp. 758–766, Dec. 2015, doi:
[262] A. F. Klaib, N. O. Alsrehin, W. Y. Melhem, H. O. Bashtawi, and 10.1109/TBCAS.2015.2504959.
A. A. Magableh, ‘‘Eye tracking algorithms, techniques, tools, and [282] P. Arpaia, N. Moccaldi, R. Prevete, I. Sannino, and A. Tedesco,
applications with an emphasis on machine learning and Internet of Things ‘‘A wearable EEG instrument for real-time frontal asymmetry monitoring
technologies,’’ Expert Syst. Appl., vol. 166, Mar. 2021, Art. no. 114037, in worker stress analysis,’’ IEEE Trans. Instrum. Meas., vol. 69, no. 10,
doi: 10.1016/j.eswa.2020.114037. pp. 8335–8343, Oct. 2020, doi: 10.1109/TIM.2020.2988744.
[263] K. I. Lee, J. H. Jeon, and B. C. Song, ‘‘Deep learning-based pupil [283] X. Zhang, J. Li, Y. Liu, Z. Zhang, Z. Wang, D. Luo, X. Zhou,
center detection for fast and accurate eye tracking system,’’ in Computer M. Zhu, W. Salman, G. Hu, and C. Wang, ‘‘Design of a fatigue
Vision—ECCV 2020 (Lecture Notes in Computer Science), vol. 12364, detection system for high-speed trains based on driver vigilance using a
A. Vedaldi, H. Bischof, T. Brox, and J.-M. Frahm, Eds. Cham, wireless wearable EEG,’’ Sensors, vol. 17, no. 3, p. 486, Mar. 2017, doi:
Switzerland: Springer, 2020, pp. 36–52, doi: 10.1007/978-3-030-58529- 10.3390/s17030486.
7_3. [284] H. Jebelli, M. M. Khalili, and S. Lee, ‘‘Mobile EEG-based workers’ stress
[264] Y. Shan, S. Li, and T. Chen, ‘‘Respiratory signal and human stress: recognition by applying deep neural network,’’ in Advances in Informatics
Non-contact detection of stress with a low-cost depth sensing camera,’’ and Computing in Civil and Construction Engineering, I.Mutis and
Int. J. Mach. Learn. Cybern., vol. 11, no. 8, pp. 1825–1837, Aug. 2020, T. Hartmann, Eds. Cham, Switzerland: Springer, 2019, pp. 173–180, doi:
doi: 10.1007/s13042-020-01074-x. 10.1007/978-3-030-00220-6_21.
[265] I. Pavlidis, J. Dowdall, N. Sun, C. Puri, J. Fei, and M. Garbey, ‘‘Interacting [285] EEG Headset Prices—An Overview of 15+ EEG Devices. Accessed:
with human physiology,’’ Comput. Vis. Image Understand., vol. 108, Sep. 20, 2022. [Online]. Available: https://imotions.com/blog/eeg-
nos. 1–2, pp. 150–170, Oct. 2007, doi: 10.1016/j.cviu.2006.11.018. headset-prices/

38242 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[286] L.-D. Liao, C.-T. Lin, K. McDowell, A. E. Wickenden, K. Gramann, [305] J. Wijsman, B. Grundlehner, J. Penders, and H. Hermens, ‘‘Trapezius
T.-P. Jung, L.-W. Ko, and J.-Y. Chang, ‘‘Biosensor technologies muscle EMG as predictor of mental stress,’’ in Wireless Health 2010.
for augmented brain–computer interfaces in the next decades,’’ New York, NY, USA: Association for Computing Machinery, Jul. 2010,
Proc. IEEE, vol. 100, no. 2, pp. 1553–1566, May 2012, doi: pp. 155–163, doi: 10.1145/1921081.1921100.
10.1109/JPROC.2012.2184829. [306] R. Luijcks, H. J. Hermens, L. Bodar, C. J. Vossen, J. V. Os,
[287] Z. Liang and M. A. Chapa Martell, ‘‘Validity of consumer activity and R. Lousberg, ‘‘Experimentally induced stress validated by EMG
wristbands and wearable EEG for measuring overall sleep parameters and activity,’’ PLoS ONE, vol. 9, no. 4, Apr. 2014, Art. no. e95215, doi:
sleep structure in free-living conditions,’’ J. Healthcare Informat. Res., 10.1371/journal.pone.0095215.
vol. 2, nos. 1–2, pp. 152–178, Jun. 2018, doi: 10.1007/s41666-018-0013- [307] D. E. Ugarte, D. Linares, G. Kemper, and C. A. Almenara, ‘‘An
1. algorithm to measure the stress level from EEG, EMG and HRV signals,’’
[288] ECG & ECHO. ECG Interpretation: Characteristics of the Normal ECG in Proc. Int. Conf. Inf. Syst. Comput. Sci. (INCISCOS), Nov. 2019,
(P-Wave, QRS Complex, ST Segment, T-Wave). Accessed: Sep. 18, 2022. pp. 346–353, doi: 10.1109/INCISCOS49368.2019.00061.
[Online]. Available: https://ecgwaves.com/topic/ecg-normal-p-wave-qrs- [308] J. Nyboer, M. M. Kreider, and L. Hannapel, ‘‘Electrical impedance
complex-st-segment-t-wave-j-point/ plethysmography,’’ Circulation, vol. 2, no. 6, pp. 811–821, Dec. 1950,
[289] V. Bhagyalakshmi, R. V. Pujeri, and G. D. Devanagavi, ‘‘GB-SVNN: doi: 10.1161/01.cir.2.6.811.
Genetic BAT assisted support vector neural network for arrhythmia [309] S. Mansouri, ‘‘Determination of cardiac output by peripheral electrical
classification using ECG signals,’’ J. King Saud Univ.-Comput. Inf. Sci., bioimpedance,’’ IEEJ Trans. Electr. Electron. Eng., vol. 15, no. 9,
vol. 33, no. 1, pp. 54–67, Jan. 2021, doi: 10.1016/j.jksuci.2018.02.005. pp. 1321–1326, Sep. 2020, doi: 10.1002/tee.23199.
[290] J. Malmivuo, ‘‘The basis of ECG diagnosis,’’ in Bioelectromagnetism: [310] A. M. Napoli, ‘‘Physiologic and clinical principles behind
Principles and Applications of Bioelectric and Biomagnetic Fields, noninvasive resuscitation techniques and cardiac output monitoring,’’
J. Malmivuo and R. Plonsey, Eds. Oxford, U.K.: Oxford Univ. Press, Cardiol. Res. Pract., vol. 2012, pp. 1–12, Jan. 2012, doi:
1995. 10.1155/2012/531908.
[291] G. Acerbi, E. Rovini, S. Betti, and A. Tirri, ‘‘A wearable system for [311] S. Aria et al., ‘‘Measuring blood pulse wave velocity with bioimpedance
stress detection through physiological data analysis,’’ in Ambient Assisted in different age groups,’’ Sensors, vol. 19, no. 4, p. 850, Feb. 2019, doi:
Living (Lecture Notes in Electrical Engineering), F. Cavallo, V. Marletta, 10.3390/s19040850.
and P. Siciliano, Eds. Cham, Switzerland: Springer, 2017, pp. 31–50, doi:
[312] T. Huynh, R. Jafari, and W.-Y. Chung, ‘‘An accurate bioimpedance
10.1007/978-3-319-54283-6_3.
measurement system for blood pressure monitoring,’’ Sensors, vol. 18,
[292] N. Isakadze and S. S. Martin, ‘‘How useful is the smartwatch ECG?’’
no. 7, p. 2095, Jun. 2018, doi: 10.3390/s18072095.
Trends Cardiovascular Med., vol. 30, no. 7, pp. 442–448, Oct. 2020, doi:
[313] D. Buxi, J.-M. Redouté, and M. R. Yuce, ‘‘Blood pressure estimation
10.1016/j.tcm.2019.10.010.
using pulse transit time from bioimpedance and continuous wave radar,’’
[293] Digit. Health Central. (2022). Best ECG Smartwatches for 2022
IEEE Trans. Biomed. Eng., vol. 64, no. 4, pp. 917–927, Apr. 2017, doi:
(FDA Approved/cleared). Accessed: Sep. 18, 2022. [Online]. Available:
10.1109/TBME.2016.2582472.
https://digitalhealthcentral.com/2021/02/23/best-ecg-smartwatch/
[314] A. M. Carek, J. Conant, A. Joshi, H. Kang, and O. T. Inan, ‘‘SeismoWatch:
[294] J. Ogorevc, G. Geršak, D. Novak, and J. Drnovšek, ‘‘Metrological evalu-
Wearable cuffless blood pressure monitoring using pulse transit time,’’
ation of skin conductance measurements,’’ Measurement, vol. 46, no. 9,
Proc. ACM Interact., Mobile, Wearable Ubiquitous Technol., vol. 1, no. 3,
pp. 2993–3001, Nov. 2013, doi: 10.1016/j.measurement.2013.06.024.
pp. 1–16, Sep. 2017, doi: 10.1145/3130905.
[295] M. D. Strauss. (2005). HandWave? Design and Manufacture
of a Wearable Wireless Skin Conductance Sensor and [315] F. Seoane, I. Mohino-Herranz, J. Ferreira, L. Alvarez, R. Buendia,
Housing. Accessed: Nov. 16, 2022. [Online]. Available: D. Ayllón, C. Llerena, and R. Gil-Pita, ‘‘Wearable biomedical mea-
https://dspace.mit.edu/handle/1721.1/32944 surement systems for assessment of mental stress of combatants in
real time,’’ Sensors, vol. 14, no. 4, pp. 7120–7141, Apr. 2014, doi:
[296] M. Winter, R. Pryss, T. Probst, and M. Reichert, ‘‘Towards the
10.3390/s140407120.
applicability of measuring the electrodermal activity in the context
of process model comprehension: Feasibility study,’’ Sensors, vol. 20, [316] M. H. Jung, K. Namkoong, Y. Lee, Y. J. Koh, K. Eom, H. Jang, W. Jung,
no. 16, p. 4561, Aug. 2020, doi: 10.3390/s20164561. J. Bae, and J. Park, ‘‘Wrist-wearable bioelectrical impedance analyzer
[297] A. Gul Airij, R. Sudirman, U. Ullah Sheikh, L. Y. Khuan, with miniature electrodes for daily obesity management,’’ Sci. Rep.,
and N. A. Zakaria, ‘‘Significance of electrodermal activity vol. 11, no. 1, pp. 1–20, Jan. 2021, doi: 10.1038/s41598-020-79667-3.
response in children with autism spectrum disorder,’’ Indonesian [317] D. Liu, G. Hong, D. Liu, and G. Hong, Wearable Electromechanical
J. Electr. Eng. Comput. Sci., vol. 19, no. 2, p. 1113, Aug. 2020, doi: Sensors and Its Applications. Rijeka, Croatia: IntechOpen, 2019, doi:
10.11591/ijeecs.v19.i2.pp1113-1120. 10.5772/intechopen.85098.
[298] C. L. Lim, C. Rennie, R. J. Barry, H. Bahramali, I. Lazzaro, B. Manor, [318] J. Mu, S. Xian, J. Yu, Z. Li, J. Zhao, J. Zhong, X. Han, X. Hou,
and E. Gordon, ‘‘Decomposing skin conductance into tonic and J. He, and X. Chou, ‘‘Flexible and wearable BaTiO3/polyacrylonitrile-
phasic components,’’ Int. J. Psychophysiol., vol. 25, no. 2, pp. 97–109, based piezoelectric sensor for human posture monitoring,’’ Sci. China
Feb. 1997, doi: 10.1016/s0167-8760(96)00713-1. Technological Sci., vol. 65, no. 4, pp. 858–869, Apr. 2022, doi:
[299] H. Kurniawan, A. V. Maslov, and M. Pechenizkiy, ‘‘Stress detection 10.1007/s11431-021-2005-0.
from speech and galvanic skin response signals,’’ in Proc. 26th IEEE [319] G. Ciuti, L. Ricotti, A. Menciassi, and P. Dario, ‘‘MEMS sensor
Int. Symp. Comput.-Based Med. Syst., Jun. 2013, pp. 209–214, doi: technologies for human centred applications in healthcare, physical
10.1109/CBMS.2013.6627790. activities, safety and environmental sensing: A review on research
[300] A. Fernandes, R. Helawar, R. Lokesh, T. Tari, and A. V. Shahapurkar, activities in Italy,’’ Sensors, vol. 15, no. 3, pp. 6441–6468, Mar. 2015,
‘‘Determination of stress using blood pressure and galvanic skin doi: 10.3390/s150306441.
response,’’ in Proc. Int. Conf. Commun. Netw. Technol., Dec. 2014, [320] G. De Pasquale and A. Somà, ‘‘Energy harvesting from human motion
pp. 165–168, doi: 10.1109/CNT.2014.7062747. with piezo fibers for the body monitoring by MEMS sensors,’’ in
[301] M. Najström and B. Jansson, ‘‘Skin conductance responses as predictor of Proc. Symp. Design, Test, Integr. Packag. MEMS/MOEMS (DTIP),
emotional responses to stressful life events,’’ Behav. Res. Therapy, vol. 45, Apr. 2013, pp. 1–6.
no. 10, pp. 2456–2463, Oct. 2007, doi: 10.1016/j.brat.2007.03.001. [321] T. Dinh, T. Nguyen, H. Phan, T. Nguyen, V. T. Dau, N. Nguyen,
[302] V. Gohel and N. Mehendale, ‘‘Review on electromyography signal and D. V. Dao, ‘‘Advances in rational design and materials of high-
acquisition and processing,’’ Biophys. Rev., vol. 12, no. 6, pp. 1361–1367, performance stretchable electromechanical sensors,’’ Small, vol. 16,
Nov. 2020, doi: 10.1007/s12551-020-00770-w. no. 14, Apr. 2020, Art. no. 1905707, doi: 10.1002/smll.201905707.
[303] Electromyography (EMG)—Mayo Clinic. Accessed: Sep. 21, 2022. [322] J. Lee, H. Espinosa, and D. James, ‘‘The inertial sensor: A base platform
[Online]. Available: https://www.mayoclinic.org/tests-procedures/ for wider adoption in sports science applications,’’ J. Fit. Res., vol. 4,
emg/about/pac-20393913 pp. 1–12, Jan. 2015.
[304] S.-E. Larsson, R. Larsson, Q. Zhang, and H. Cai, ‘‘Effects of psychophys- [323] T. Giorgino, P. Tormene, F. Lorussi, D. De Rossi, and S. Quaglini, ‘‘Sen-
iological stress on trapezius muscles blood flow and electromyography sor evaluation for wearable strain gauges in neurological rehabilitation,’’
during static load,’’ Eur. J. Appl. Physiol. Occupational Physiol., vol. 71, IEEE Trans. Neural Syst. Rehabil. Eng., vol. 17, no. 4, pp. 409–415,
no. 6, pp. 493–498, Nov. 1995, doi: 10.1007/bf00238550. Aug. 2009, doi: 10.1109/TNSRE.2009.2019584.

VOLUME 12, 2024 38243


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[324] M. M. Rodgers, V. M. Pai, and R. S. Conroy, ‘‘Recent advances in [345] Empatica | Medical Devices, AI and Algorithms for Remote
wearable sensors for health monitoring,’’ IEEE Sensors J., vol. 15, no. 6, Patient Monitoring. Accessed: Mar. 20, 2023. [Online]. Available:
pp. 3119–3126, Jun. 2015, doi: 10.1109/JSEN.2014.2357257. https://www.empatica.com/
[325] D. K. Shaeffer, ‘‘MEMS inertial sensors: A tutorial overview,’’ [346] Neurotech for the Global Community. Accessed: Mar. 20, 2023. [Online].
IEEE Commun. Mag., vol. 51, no. 4, pp. 100–109, Apr. 2013, doi: Available: https://www.emotiv.com/
10.1109/MCOM.2013.6495768. [347] Advanced Health & Fitness Smartwatch | Shop Fitbit
[326] M. Etemadi and O. T. Inan, ‘‘Wearable ballistocardiogram and seismocar- Sense 2. Accessed: Mar. 20, 2023. [Online]. Available:
diogram systems for health and performance,’’ J. Appl. Physiol., vol. 124, https://www.fitbit.com/global/us/products/smartwatches/sense2
no. 2, pp. 452–461, Feb. 2018, doi: 10.1152/japplphysiol.00298.2017.
[348] Flowtime Biosensing Meditation Headband. Accessed: Mar. 20, 2023.
[327] L. Giovangrandi, O. T. Inan, R. M. Wiard, M. Etemadi, and
[Online]. Available: https://www.meetflowtime.com/products/flowtime-
G. T. A. Kovacs, ‘‘Ballistocardiography—A method worth revisiting,’’
biosensing-meditation-headband
in Proc. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc., Aug. 2011,
pp. 4279–4282, doi: 10.1109/IEMBS.2011.6091062. [349] GlucoMen Areo and Areo 2K. Accessed: Mar. 20, 2023. [Online].
[328] D. Rai, H. K. Thakkar, S. S. Rajput, J. Santamaria, C. Bhatt, and F. Roca, Available: https://www.glucomen.co.uk/areo/
‘‘A comprehensive review on seismocardiogram: Current advancements [350] GraphWear. Accessed: Mar. 20, 2023. [Online]. Available:
on acquisition, annotation, and applications,’’ Mathematics, vol. 9, no. 18, https://www.graphwear.co
p. 2243, Sep. 2021, doi: 10.3390/math9182243. [351] Vital Patch. Accessed: Mar. 20, 2023. [Online]. Available:
[329] A. Taebi, B. Solar, A. Bomar, R. Sandler, and H. Mansy, ‘‘Recent https://vitalconnect.com/
advances in seismocardiography,’’ Vibration, vol. 2, no. 1, pp. 64–86, [352] Awario—AI Powered Heart Arrhythmia Detection System. Accessed:
Jan. 2019, doi: 10.3390/vibration2010005. Mar. 20, 2023. [Online]. Available: https://awario.global/
[330] G. Fierro, F. Silveira, and R. Armentano, ‘‘Central blood pressure [353] Xiaomi Hipee Posture Correction Device: Where to Buy, Features
monitoring method oriented to wearable devices,’’ Health Technol., vol. 6, and Reviews. Accessed: Mar. 20, 2023.[Online]. Available:
no. 3, pp. 197–204, Nov. 2016, doi: 10.1007/s12553-016-0149-z. https://xiaomipedia.com/en/p/hipee-posture-correction-device/
[331] A. D. Wiens, M. Etemadi, S. Roy, L. Klein, and O. T. Inan, [354] Wearable ECG Necklace With Bluetooth
‘‘Toward continuous, noninvasive assessment of ventricular Connection. Accessed: Mar. 20, 2023. [Online]. Available:
function and hemodynamics: Wearable ballistocardiography,’’ IEEE https://www.ecgnecklace.com/wearable-ecg-physical-activity-monitor/
J. Biomed. Health Informat., vol. 19, no. 4, pp. 1435–1442, Jul. 2015,
doi: 10.1109/JBHI.2014.2359937. [355] MuseTM—Meditation Made Easy. Accessed: Mar. 20, 2023. [Online].
[332] P. Yousefian, S. Shin, A. S. Mousavi, A. Tivay, C.-S. Kim, R. Mukkamala, Available: https://choosemuse.com/muse-2/
D.-G. Jang, B. H. Ko, J. Lee, U.-K. Kwon, Y. H. Kim, and [356] K’Watch Glucose. Accessed: Mar. 20, 2023. [Online]. Available:
J.-O. Hahn, ‘‘Pulse transit time-pulse wave analysis fusion based https://www.pkvitality.com/ktrack-glucose/
on wearable wrist ballistocardiogram for cuff-less blood pressure [357] Lab-on-SkinTM Technology—Xsensio | Lab-on-SkinTM Sensing
trend tracking,’’ IEEE Access, vol. 8, pp. 138077–138087, 2020, doi: Platform. Accessed: Mar. 20, 2023. [Online]. Available:
10.1109/ACCESS.2020.3012384. https://xsensio.com/lab-on-skin-technology/
[333] H. Park, W. Park, and C. H. Lee, ‘‘Electrochemically active materials and [358] (3441). What Can I Still Do With My Microsoft Band?—Microsoft
wearable biosensors for the in situ analysis of body fluids for human Support. Accessed: Mar. 20, 2023. [Online]. Available:
healthcare,’’ NPG Asia Mater., vol. 13, no. 1, p. 23, Dec. 2021, doi: https://support.microsoft.com/en-us/topic/what-can-i-still-do-with-
10.1038/s41427-020-00280-x. my-microsoft-band-a2a59355-5be0-3441-9fff-4dc27bcbafb5
[334] R. S. P. Malon, S. Sadir, M. Balakrishnan, and E. P. Córcoles, [359] ESense Skin Response—Mobile Biofeedback With Your Smartphone or
‘‘Saliva-based biosensors: Noninvasive monitoring tool for clinical Tablet. Accessed: Mar. 20, 2023. [Online]. Available: https://mindfield-
diagnostics,’’ BioMed Res. Int., vol. 2014, pp. 1–20, Feb. 2014, doi: esense.com/esense-skin-response/
10.1155/2014/962903.
[360] MindWave. Accessed: Mar. 20, 2023. [Online]. Available:
[335] C. E. Lin, Y. Ito, A. Deng, J. Johns, D. Matloff, C. B. Cook,
https://store.neurosky.com/pages/mindwave
K. Sode, and J. T. La Belle, ‘‘A disposable tear glucose biosensor—
Part 5: Improvements in reagents and tear sampling component,’’ [361] Wireless Wearable Sensor For Medical And Sports Use. Accessed:
J. Diabetes Sci. Technol., vol. 12, no. 4, pp. 842–846, Apr. 2018, doi: Mar. 20, 2023. [Online]. Available: https://www.movesense.com/
10.1177/1932296818769944. [362] NIRSport2 | FNIRS Systems | NIRS Devices | NIRx. Accessed: Mar. 20,
[336] M. Chung, G. Fortunato, and N. Radacsi, ‘‘Wearable flexible sweat sen- 2023. [Online]. Available: https://nirx.net/nirsport
sors for healthcare monitoring: A review,’’ J. Roy. Soc. Interface, vol. 16, [363] Wellue O2Ring Smart Ring Pulse Oximeter. Overnight
no. 159, Oct. 2019, Art. no. 20190217, doi: 10.1098/rsif.2019.0217. Tracking. Accessed: Mar. 20, 2023. [Online]. Available:
[337] G. Li and D. Wen, ‘‘Wearable biochemical sensors for human https://getwellue.com/pages/o2ring-oxygen-monitor
health monitoring: Sensing materials and manufacturing technologies,’’ [364] Oura Ring: Accurate Health Information Accessible To
J. Mater. Chem. B, vol. 8, no. 16, pp. 3423–3436, Apr. 2020, doi: Everyone. Accessed: Mar. 20, 2023. [Online]. Available:
10.1039/c9tb02474c. https://ouraring.com
[338] V. Naresh and N. Lee, ‘‘A review on biosensors and recent development [365] Prana Wearable. Accessed: Mar. 20, 2023. [Online]. Available:
of nanostructured materials-enabled biosensors,’’ Sensors, vol. 21, no. 4, https://prana.co/
p. 1109, Feb. 2021, doi: 10.3390/s21041109.
[366] Digital Precision Medicine for Mental Health. Accessed: Mar. 20, 2023.
[339] S. Coyle, V. F. Curto, F. Benito-Lopez, L. Florea, and D. Diamond,
[Online]. Available: https://www.feeltherapeutics.com/
‘‘Wearable bio and chemical sensors,’’ in Wearable Sensors, E. Sazonov
and M. R. Neuman, Eds. New York, NY, USA: Academic, 2014, [367] Shimmer Wearable Sensor Technology | Wireless IMU | ECG
pp. 65–83, doi: 10.1016/B978-0-12-418662-0.00002-7. | EMG | GSR. Accessed: Mar. 20, 2023. [Online]. Available:
[340] J. Kim, A. S. Campbell, B. E.-F. de Ávila, and J. Wang, ‘‘Wearable https://shimmersensing.com/
biosensors for healthcare monitoring,’’ Nature Biotechnol., vol. 37, no. 4, [368] M. Strauss, C. Reynolds, S. Hughes, K. Park, G. McDarby, and
pp. 389–406, Apr. 2019, doi: 10.1038/s41587-019-0045-y. R. W. Picard, ‘‘The handwave Bluetooth skin conductance sensor,’’
[341] AutoSense: A Wireless Sensor System To Quantify Personal Expo- in Affective Computing and Intelligent Interaction (Lecture Notes in
sures to Psychosocial Stress and Addictive Substances in Natu- Computer Science), J. Tao, T. Tan, and R. W. Picard, Eds. Berlin,
ral Environments. Accessed: Mar. 20, 2023. [Online]. Available: Germany: Springer, 2005, pp. 699–706, doi: 10.1007/11573548_90.
https://sites.google.com/site/autosenseproject/ [369] ZephyrTM Performance Systems | Performance Monitoring
[342] Biobeat Medical—Smart Vital Signs Monitoring. Accessed: Technology. Accessed: Mar. 20, 2023. [Online]. Available:
Mar. 20, 2023. [Online]. Available: https://www.bio-beat.com/ https://www.zephyranywhere.com/
[343] VitalStream by Caretaker Medical. Accessed: Mar. 20, 2023. [Online]. [370] Stress Detection by Machine Learning and Wearable
Available: https://caretakermedical.net/ Sensors | 26th International Conference on Intelligent User
[344] Dexcom Continuous Glucose Monitoring. Accessed: Mar. 20, 2023. Interfaces—Companion. Accessed: Dec. 1, 2022. [Online]. Available:
[Online]. Available: https://www.dexcom.com https://dl.acm.org/doi/10.1145/3397482.3450732

38244 VOLUME 12, 2024


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

[371] S. S. Panicker and P. Gayathri, ‘‘A survey of machine learning [390] Human Stress And Pressure Detection Using A Machine Learning
techniques in physiology based mental stress detection systems,’’ Approach With Eeg Signal. Accessed: Feb. 1, 2023. [Online]. Available:
Biocybern. Biomed. Eng., vol. 39, no. 2, pp. 444–469, Apr. 2019, doi: https://www.webology.org/abstract.php?id=2857
10.1016/j.bbe.2019.01.004. [391] R. Setiawan, F. Budiman, and W. I. Basori, ‘‘Stress diagnostic
[372] S. Gedam and S. Paul, ‘‘Automatic stress detection using wearable system and digital medical record based on Internet of Things,’’ in
sensors and machine learning: A review,’’ in Proc. 11th Int. Conf. Com- Proc. Int. Seminar Intell. Technol. Appl., Aug. 2019, pp. 348–353, doi:
put., Commun. Netw. Technol. (ICCCNT), Jul. 2020, pp. 1–7, doi: 10.1109/ISITIA.2019.8937273.
10.1109/ICCCNT49239.2020.9225692. [392] A. Ceniza and A. Cantara, ‘‘Stress sensor prototype: Determining
[373] S. Gedam and S. Paul, ‘‘A review on mental stress detection the stress level in using a computer through validated self-made
using wearable sensors and machine learning techniques,’’ IEEE heart rate (HR) and galvanic skin response (GSR) sensors and fuzzy
Access, vol. 9, pp. 84045–84066, 2021, doi: 10.1109/ACCESS.2021. logic algorithm,’’ Int. J. Eng. Res. Technol., vol. 5, no. 3, pp. 1–10,
3085502. Mar. 2016. Accessed: Feb. 1, 2023.
[374] D. Cho, J. Ham, J. Oh, J. Park, S. Kim, N.-K. Lee, and B. Lee, ‘‘Detection [393] S. Naqvi, A. Z. Shaikh, T. Altaf, and S. Singh, ‘‘Fuzzy logic enabled
of stress levels from biosignals measured in virtual reality environments stress detection using physiological signals,’’ in Emerging Technologies
using a kernel-based extreme learning machine,’’ Sensors, vol. 17, no. 10, in Computing (Lecture Notes of the Institute for Computer Sciences),
p. 2435, Oct. 2017, doi: 10.3390/s17102435. M. H. Miraz, G. Southall, M. Ali, A. Ware, and S. Soomro, Eds. Cham,
Switzerland: Springer, 2021, pp. 161–173, doi: 10.1007/978-3-030-
[375] U. Bhushan and S. Maji, ‘‘Prediction and analysis of stress using machine
90016-8_11.
learning: A review,’’ in 3rd Doctoral Symp. Comput. Intell. (Lecture Notes
[394] A. G. Airij, R. Sudirman, and U. U. Sheikh, ‘‘GSM and GPS based
in Networks and Systems), A. Khanna, D. Gupta, V. Kansal, G. Fortino,
real-time remote physiological signals monitoring and stress levels
and A. E. Hassanien, Eds. Singapore: Springer, 2023, pp. 419–432, doi:
classification,’’ in Proc. 2nd Int. Conf. BioSignal Anal., Process.
10.1007/978-981-19-3148-2_35.
Syst. (ICBAPS), Jul. 2018, pp. 130–135, doi: 10.1109/ICBAPS.2018.
[376] S. Beniczky, P. Karoly, E. Nurse, P. Ryvlin, and M. Cook, ‘‘Machine 8527406.
learning and wearable devices of the future,’’ Epilepsia, vol. 62, no. S2, [395] N. Keshan, P. V. Parimi, and I. Bichindaritz, ‘‘Machine learning for
Mar. 2021, Art. no. S2, doi: 10.1111/epi.16555. stress detection from ECG signals in automobile drivers,’’ in Proc. IEEE
[377] P. Bobade and M. Vani, ‘‘Stress detection with machine learning and Int. Conf. Big Data (Big Data), Oct. 2015, pp. 2661–2669, doi:
deep learning using multimodal physiological data,’’ in Proc. 2nd 10.1109/BIGDATA.2015.7364066.
Int. Conf. Inventive Res. Comput. Appl. (ICIRCA), Jul. 2020, pp. 51–57, [396] A. P. Cruz, A. Pradeep, K. R. Sivasankar, and K. S. Krishnaveni,
doi: 10.1109/ICIRCA48905.2020.9183244. ‘‘A decision tree optimised SVM model for stress detection using
[378] O. Faust, Y. Hagiwara, T. J. Hong, O. S. Lih, and U. R. Acharya, ‘‘Deep biosignals,’’ in Proc. Int. Conf. Commun. Signal Process. (ICCSP),
learning for healthcare applications based on physiological signals: Jul. 2020, pp. 0841–0845, doi: 10.1109/ICCSP48568.2020.
A review,’’ Comput. Methods Programs Biomed., vol. 161, pp. 1–13, 9182043.
Jul. 2018, doi: 10.1016/j.cmpb.2018.04.005. [397] A. Sano and R. W. Picard, ‘‘Stress recognition using
[379] L. Han, Q. Zhang, X. Chen, Q. Zhan, T. Yang, and Z. Zhao, wearable sensors and mobile phones,’’ in Proc. Humaine
‘‘Detecting work-related stress with a wearable device,’’ Assoc. Conf. Affect. Comput. Intell. Interact., Sep. 2013, pp. 671–676,
Comput. Ind., vol. 90, pp. 42–49, Sep. 2017, doi: 10.1016/j.compind. doi: 10.1109/ACII.2013.117.
2017.05.004. [398] Y. S. Can, N. Chalabianloo, D. Ekiz, J. Fernandez-Alvarez, G. Riva,
[380] V. Rajendra Mali and P. A. R. Surve, ‘‘Stress prediction using random and C. Ersoy, ‘‘Personal stress-level clustering and decision-level
forest linear regression,’’ Int. J. Eng. Appl. Sci. Technol., vol. 6, no. 3, smoothing to enhance the performance of ambulatory stress detection
pp. 1–15, Jul. 2021, doi: 10.33564/ijeast.2021.v06i03.048. with smartwatches,’’ IEEE Access, vol. 8, pp. 38146–38163, 2020, doi:
[381] T. Lim, J. Heo, K. Jeong, and H. Ghim, ‘‘Stress assesment based on bio- 10.1109/ACCESS.2020.2975351.
signals using random forest algorithm,’’ J. Korean Soc. Saf., vol. 35, no. 1, [399] M. Gil-Martin, R. San-Segundo, A. Mateos, and J. Ferreiros-Lopez,
pp. 62–69, 2020, doi: 10.14346/JKOSOS.2020.35.1.62. ‘‘Human stress detection with wearable sensors using convolutional
neural networks,’’ IEEE Aerosp. Electron. Syst. Mag., vol. 37, no. 1,
[382] J. Kim, D. Yoon, and H.-S. Kim, ‘‘Random forest approach in prediction
pp. 60–70, Jan. 2022, doi: 10.1109/MAES.2021.3115198.
workers’ stress from personality traits,’’ in Proc. 13th Int. Conf. Inf. Com-
mun. Technol. Converg. (ICTC), Oct. 2022, pp. 2356–2358, doi: [400] H. Lin, J. Jia, Q. Guo, Y. Xue, J. Huang, L. Cai, and L. Feng,
10.1109/ICTC55196.2022.9952879. ‘‘Psychological stress detection from cross-media microblog data
using deep sparse neural network,’’ in Proc. IEEE Int. Conf. Mul-
[383] R. Jegan, S. Mathuranjani, and P. Sherly, ‘‘Mental stress detection
timedia Expo (ICME), Jul. 2014, pp. 1–6, doi: 10.1109/ICME.2014.
and classification using SVM classifier: A pilot study,’’ in Proc. 6th
6890213.
Int. Conf. Devices, Circuits Syst. (ICDCS), Apr. 2022, pp. 139–143, doi:
[401] M. Chauhan, S. V. Vora, and D. Dabhi, ‘‘Effective stress detection using
10.1109/ICDCS54290.2022.9780795.
physiological parameters,’’ in Proc. Int. Conf. Innov. Inf., Embedded
[384] V. Dham, K. Rai, and U. Soni, ‘‘Mental stress detection using artificial
Commun. Syst. (ICIIECS), Mar. 2017, pp. 1–6, doi: 10.1109/ICI-
intelligence models,’’ J. Physics: Conf. Ser., vol. 1950, no. 1, Aug. 2021,
IECS.2017.8275853.
Art. no. 012047, doi: 10.1088/1742-6596/1950/1/012047.
[402] B. Blobel, PHealth 2021: Proceedings of the 18th International
[385] R. Ahuja and A. Banga, ‘‘Mental stress detection in university students Conference on Wearable Micro and Nano Technologies for Personalized
using machine learning algorithms,’’ Proc. Comput. Sci., vol. 152, Health. Amsterdam, The Netherlands: IOS Press, 2021.
pp. 349–353, Jan. 2019, doi: 10.1016/j.procs.2019.05.007. [403] M. Gjoreski, M. Luštrek, M. Gams, and H. Gjoreski, ‘‘Monitoring
[386] A. Ghaderi, J. Frounchi, and A. Farnam, ‘‘Machine learning-based stress with a wrist device using context,’’ J. Biomed. Informat., vol. 73,
signal processing using physiological signals for stress detection,’’ in pp. 159–170, Sep. 2017, doi: 10.1016/j.jbi.2017.08.006.
Proc. 22nd Iranian Conf. Biomed. Eng. (ICBME), Nov. 2015, pp. 93–98, [404] T. A. Roldán-Rojo, E. Rendón-Veléz, and S. Carrizosa, ‘‘Stressors
doi: 10.1109/ICBME.2015.7404123. and algorithms used for stress detection: A review,’’ in Proc. 9th
[387] R. Gupta, M. A. Alam, and P. Agarwal, ‘‘Modified support Int. Conf. Affect. Comput. Intell. Interact. (ACII), Sep. 2021, pp. 1–8, doi:
vector machine for detecting stress level using EEG signals,’’ 10.1109/ACII52823.2021.9597456.
Comput. Intell. Neurosci., vol. 2020, pp. 1–14, Aug. 2020, doi: [405] C. Maaoui, F. Bousefsaf, and A. Pruski, ‘‘Automatic human stress
10.1155/2020/8860841. detection based on WebCam photoplethysmographic signals,’’
[388] Y. S. Can, N. Chalabianloo, D. Ekiz, and C. Ersoy, ‘‘Continuous stress J. Mech. Med. Biol., vol. 16, no. 4, Jun. 2016, Art. no. 1650039,
detection using wearable sensors in real life: Algorithmic programming doi: 10.1142/s0219519416500391.
contest case study,’’ Sensors, vol. 19, no. 8, p. 1849, Apr. 2019, doi: [406] M. Sugiyama, Introduction to Statistical Machine Learning. San Mateo,
10.3390/s19081849. CA, USA: Morgan Kaufmann, 2015.
[389] T. Rahman, A. Ghosh, and Md. M. H. Shuvo, ‘‘Mental stress recog- [407] A. Arsalan, M. Majid, I. Fareed Nizami, W. Manzoor, S. Muhammad
nition using K-nearest neighbor (KNN) classifier on EEG signals,’’ Anwar, and J. Ryu, ‘‘Human stress assessment: A comprehensive review
in Proc. Int. Conf. Mater., Electron. Inf. Eng. (ICMEIE), Jun. 2015, of methods using wearable sensors and non-wearable techniques,’’ 2022,
pp. 1–4. arXiv:2202.03033.

VOLUME 12, 2024 38245


G. Taskasaplidis et al.: Review of Stress Detection Methods Using Wearable Sensors

GEORGIOS TASKASAPLIDIS was born in PANAGIOTIS D. BAMIDIS (Member, IEEE) was


Gütersloh, North Rhine-Westphalia, Germany. born in Tihero, Evrou, Greece. He received the
He received the B.Sc. degree in electronics Diploma degree in physics from the Aristotle
engineering from the Technological Educational University of Thessaloniki (AUTH), Thessaloniki,
Institute of Lamia, Greece, in 1995, and the M.Sc. Greece, in 1990, the M.Sc. degree (Hons.) in
degree in information technology (IT) from the medical physics from the University of Surrey,
University of the West of Scotland, U.K., in 2007. Guildford, U.K., in 1992, and the Ph.D. degree
He is currently pursuing the Ph.D. degree with the in bioelectromagnetism and functional brain anal-
Microprocessors—Microcontrollers, Automatic ysis and imaging from Open University, Milton
Systems Control and Robotics Laboratory, Infor- Keynes, U.K., in 1996.
matics Department, University of Western Macedonia (UOWM), Kastoria, In 2017, he became a Visiting Professor in medical education technology,
Greece. innovation and change with the Leeds Institute of Medical Education
Since 2016, he has been with the Laboratory Teaching Personnel, (LIME), University of Leeds, U.K. Since 2020, he has been leading the
Informatics Department, UOWM. He worked for several years for Siemens Medical Education Innovation & Research Unit (MEIRU) and the Special
as an Electronics Engineer, before becoming a member of UOWM. His Unit for Biomedical Research and Education (SUBRE), School of Medicine.
research interests include bioscience, biomarkers, and data analysis with He is currently a Professor in medical physics, informatics, and medical
informatics. education and the Director of the Laboratory of Medical Physics and Digital
Innovation, School of Medicine, AUTH. He is also the coordinator of ten
DIMITRIS A. FOTIADIS was born in Edessa, large European projects and the principal investigator for many national and
Greece. He received the B.Sc. degree in electronics international funded projects.
engineering from the Technological Educational Dr. Bamidis is also the President of the Hellenic Biomedical Technology
Institute of Thessaloniki, Greece, in 1996, the Society, HL7 Hellas, the International Society of Applied Neuroscience,
M.Sc. degree in electronics (digital systems) from a member of the Administration Boards of Other Societies and Patient
the University of Hertfordshire, U.K., in 1997, Associations, the Chairperson/Organizer of some 20 international confer-
and the Ph.D. degree in medical informatics from ences, and several National Biomedical Technology Conferences. Since
the Medical Informatics Laboratory, School of 2012, he has been established LLM Care Ecosystem (www.llmcare.gr), the
Medicine, Aristotle University of Thessaloniki Business Exploitation of the LLM Project, which is a two-star reference
(AUTH), Greece, in 2016. site of the EIP-on-AHA. In 2013, he established the Active and Healthy
Since 2008, he has been an Assistant Professor with the Informatics Ageing Living Laboratory, Thessaloniki (ThessAHALL) (http://www.aha-
Department, University of Western Macedonia (UOWM), Kastoria, Greece, livinglabs.com/) which, in 2018, he became a full member of the
where he is currently the Director of the Lifelong Education Training European Network of Living Laboratories (ENoLL). He received Prizes
Center and the Director of the Mircoprocessors/Microcontrollers, Automated for the Best Track Record in funded research projects (AUTH Research
Control Electronic Systems and Robotics Scientific Laboratory, Informatics Committee, in 2009, AUTH Dean of Health Sciences, in 2016, and AUTH
Department. He is the author of two books, several journals, and international Excellence Committee, in 2021) and the Best Overall High/Extra-Ordinary
conference papers; and is the owner of a national patent. His research Academic Performance (AUTH Dean of Health Sciences, in 2018, 2019,
interests include biomedical engineering, microcontrollers, microprocessors, and 2020).
and robotic system design and implementation.

38246 VOLUME 12, 2024

You might also like