Applications of wearable sensors in upper extremity MSK conditions: a scoping review

Abstract

Purpose

This scoping review uniquely aims to map the current state of the literature on the applications of wearable sensors in people with or at risk of developing upper extremity musculoskeletal (UE-MSK) conditions, considering that MSK conditions or disorders have the highest rate of prevalence among other types of conditions or disorders that contribute to the need for rehabilitation services.

Materials and methods

The preferred reporting items for systematic reviews and meta-analysis (PRISMA) extension for scoping reviews guideline was followed in this scoping review. Two independent authors conducted a systematic search of four databases, including PubMed, Embase, Scopus, and IEEEXplore. We included studies that have applied wearable sensors on people with or at risk of developing UE-MSK condition published after 2010. We extracted study designs, aims, number of participants, sensor placement locations, sensor types, and number, and outcome(s) of interest from the included studies. The overall findings of our scoping review are presented in tables and diagrams to map an overview of the existing applications.

Results

The final review encompassed 80 studies categorized into clinical population (31 studies), workers’ population (31 studies), and general wearable design/performance studies (18 studies). Most were observational, with 2 RCTs in workers’ studies. Clinical studies focused on UE-MSK conditions like rotator cuff tear and arthritis. Workers’ studies involved industrial workers, surgeons, farmers, and at-risk healthy individuals. Wearable sensors were utilized for objective motion assessment, home-based rehabilitation monitoring, daily activity recording, physical risk characterization, and ergonomic assessments. IMU sensors were prevalent in designs (84%), with a minority including sEMG sensors (16%). Assessment applications dominated (80%), while treatment-focused studies constituted 20%. Home-based applicability was noted in 21% of the studies.

Conclusion

Wearable sensor technologies have been increasingly applied to the health care field. These applications include clinical assessments, home-based treatments of MSK disorders, and monitoring of workers’ population in non-standardized areas such as work environments. Assessment-focused studies predominate over treatment studies. Additionally, wearable sensor designs predominantly use IMU sensors, with a subset of studies incorporating sEMG and other sensor types in wearable platforms to capture muscle activity and inertial data for the assessment or rehabilitation of MSK conditions.

Introduction

Wearable motion sensors are light, non-invasive electronic devices [1] that can be comfortably worn or carried by individuals. These systems are designed to assess, monitor, and report specific physical, positional, movement or physiological parameters. They can be used to monitor people providing information arising from different body segments such as the wrist, shoulder, back, chest, knees, and other limbs. Through technological advancements in recent years, various applications of wearable sensors have been suggested and proposed in different fields such as engineering [2], health care [3], gaming [4], daily activity [5], social networking [6], and the military [7]. One of the main applications of wearable sensors in health care studies is the assessment of motion or muscle activity, in patients with various disorders or diseases or healthy individuals. These wearable sensors enable clinicians to have reduced assessment times and obtain objective and quantifiable data of individuals. Sensors may complement the subjective health measurement outcomes derived from clinicians’ and patients’ perspective since they assess different parameters. Moreover, unobtrusive, and continuous measurements can be recorded by using wearable sensors. This enables functional or remote monitoring of patient status in real world daily life activities. Application of this data can support implementation of more affordable or more customized therapy for patients in their home environment, allowing a better and more extensive rehabilitation process [8]. Finally, corrective feedback can be presented to patients by increasing their understanding of performing correct movement patterns during therapy sessions (in the clinic or at home) [9].

Inertial measurement units (IMUs) and surface electromyography sensors (sEMG) are the two leading technologies that are used for the measurement of movement quality across various joints [10, 11]. An IMU usually contains an accelerometer, a gyroscope, and a magnetometer that provides linear accelerations, angular velocities, and strength of magnetic fields in 3 dimensions, respectively [12]. Thus, it can be utilized to assess the body segments’ motions. On the other hand, the sEMG sensor provides a time-stamped signal, frequency, and strength of muscle activity [13]. Therefore, it can obtain more complementary information regarding motor function to therapists and clinicians [8]. Additionally, other types of wearable sensors such as potentiometers and encoders [14, 15] and piezoresistive sensors [16] translate the resistance alterations made by bending sensors or the presence or absence of light made by angular displacement to angles or other motion-related outcomes. However, the applicability of IMUs sensors is increasing rapidly compared to these sensors.

One of the main fields of health care that requires continuous, reliable, and objective clinical measurements or rehabilitation is Musculoskeletal (MSK) disorders. In this regard, 1.71 billion people worldwide have musculoskeletal conditions, which is the highest rate among other types of conditions or disorders that contribute to the need for rehabilitation services [17]. Specifically, Upper extremity musculoskeletal (UE-MSK) disorders and conditions are a significant concern in today’s world since they impose health burdens on patients and a substantial economic burden on society due to sick leave and health care expenses [18]. Although a well-founded and reliable specific global prevalence rate for UE-MSK conditions around the world has not been obtained (due to a lack of a globally accepted definition of UE-MSK disorders or conditions), a conducted study by Huisstede et al. demonstrated that a significant proportion of the MSK disorder population could fit in this category [18]. Therefore, improving the quality of health care, evaluation, and rehabilitation of UE-MSK conditions or disorders will significantly assist the clinicians and patients. In this regard, providing objective, easy-to-use, cheap, and rapid measurements of motion quality can substantially increase the rehabilitation accuracy and diagnosis efficiency and reduce the related costs of therapy sessions. Wearable sensors can be a perfect choice to be applied for assessments of UE-MSK disorders since they can bring forth the mentioned advantages. On the other hand, recent quarantines and health guidelines made by governments and health organizations due to the Covid-19 outbreak worldwide have significantly reduced in-person rehabilitation sessions or therapy visits. This further necessitates the use of wearable sensors for remote and at-home assessments of movement or motion quality.

To date, several reviews have been conducted to assess various types and applications of wearable sensors in rehabilitation or other healthcare fields and evaluate the functionality of these applications. A recent scoping review by Kim et al. has provided the applications of wearable sensors for assessment and treatment of upper extremities (UE) in the population of stroke patients [19]. The reviewed articles in this study applied wearable sensors to obtain UE functional motion, sort motor impairment/activity limitation, augment UE training by providing various types of feedback. Moreover, this review has demonstrated the application of wearable sensors in determining the home-based rehabilitation dosages, the characterization of daily UE use patterns in individuals’ lives, and the rate of adherence to home-based therapy sessions [19]. In another literature review, a similar population and context have been examined by Maceira-Elvira et al. [8]. This literature review has aimed to present an overview of applications of wearable sensors in stroke upper extremity rehabilitation research from different aspects. The study assessed the different wearable sensor technologies in the stroke population, their data processing methods, and instruments. Finally, it has been concluded that aside from the advantages of wearable sensors, IMUs and sEMG sensors offer the best aspects of unobtrusiveness, robustness, user-friendliness, and data quality [8].

Another scoping review study by Sethi et al. has focused on articles investigating the use of inertial motion sensors, sEMG-based, and e-textiles-based interactive wearable technologies [9]. This review summarizes the current applications, limitations, and future of inertial motion and sEMG sensors on different populations such as healthy individuals, stroke patients and neurologically impaired groups. However, it has been mentioned that wearable sensor technologies have encountered certain limitations such as large size equipment, a limited utility for clinical applications, and burdensome setup processes. On the other hand, it has been prospected that through the growth of cloud systems and machine learning algorithms, the data transfer process of these systems will become more convenient [9]. The biofeedback designs for home-based rehabilitation applications have been examined in another scoping review. In this regard, it has been reported that the analyzed feedback introduced in the studies were mainly based on visual, concurrent, and descriptive representations. Moreover, the included articles have investigated the potential reasons for using a feedback system, its user-friendliness, and evaluations [20]. In total, the mentioned reviews have extracted specific parameters of the included articles, such as the number of participants, data collection procedure, sensor type, number, and placement location, data processing methods, assessment type, and results.

Wearable health systems have been reviewed to be applied in clinical practice by Lu et al. [21]. This article has claimed that wearable medical devices have been applied to all parts and limbs of the human body. Furthermore, it has categorized the devices into four application areas: health and safety monitoring, chronic disease management, disease diagnosis/treatment, and rehabilitation. Similarly, this study declares limitations such as the absence of user-friendly solutions, security and privacy concerns, and the lack of industry standards [21].

All these reviews have aimed to provide a map of all available evidence for their research question and highlight the existing gaps in the examined contexts and populations, such as stroke patients or home-based rehabilitation. In contrast to systematic reviews, none of these scoping reviews have conducted a critical appraisal of individual sources of evidence since this step is considered optional for scoping reviews based on PRISMA extension for scoping reviews (PRISMA-ScR checklist and explanation) [22].

Considering the advantages presented by the using wearable sensors in the context of UE-MSK clinical applications, along with the distinct applications of wearable sensor systems in UE-MSK conditions, and the challenges associated with their functionality and usability, this scoping review aims to offer an extensive overview of the most recent findings in related studies. Additionally, we intend to shed light on the prevalent challenges and identify existing gaps within this research field.

In summary, the novelties of our review were covering a broad spectrum of wearable sensor applications in individuals with or at risk of developing UE-MSK conditions, encompassing diverse populations and applications, and categorizing studies into clinical population, workers’ population, and general wearable design/performance studies.

Materials and methods

In the present scoping review, the outlined guidelines suggested by the preferred reporting items for systematic reviews and meta-analysis (PRISMA) extension for scoping reviews have been considered and applied [22]. The population, concept, context (PCC) structure was selected to identify the critical elements of the research question of this study for conceptualizing purposes. In this regard, the focused populations are patients and individuals who are either diagnosed with UE-MSK disorders (e.g., arthritis or carpal tunnel syndrome patients) or individuals who are at risk of developing UE-MSK conditions (e.g., manufacturing workers, surgeons, farmers), UE wearable sensors (concept), and assessment or rehabilitation (context) [22, 23]. The scoping reviews cannot be registered within the International Prospective Register of Systematic Reviews; however, it has been registered on the Open Science Framework (https://osf.io/8h2mn/).

Databases and systematic search

Two independent authors (SM and ES) conducted a systematic search of PubMed, Embase, Scopus, and IEEEXplore databases on 4 January 2023. Three categories of keywords and their iterations were used to obtain the relevant articles. The first group of keywords included wearable sensors concept and its iterations such as wearable sensor, wearable electrode, wearable electronic, wearable device, smart prosthesis, electronic textile, IMU, and inertial measurement unit. IMU and its iterations were included in this group since it is the most applied sensor for motion measurements; however, some studies have exploited IMUs and applied a wearable platform but have not mentioned the explicit iterations of the wearable sensor in their script. The second group includes upper extremity, upper limb, shoulder, hand, wrist, elbow, arm, forearm, and finger to obtain the upper limb related articles, and the third group includes the keywords related to the musculoskeletal system such as musculoskeletal, MSK, MSD, muscle, and bone. In Additional file 1: Appendix A, the detailed applied search strategies for each database have been presented. Moreover, the references of included full-text reviewed articles and Google Scholar were examined to search for additional relevant articles. Regarding searching the grey literature, conference abstracts were examined and screened to be included. In case of any disagreement between the two authors regarding the inclusion of a study in the review, the third author (JM) helped to resolve the disagreement.

Selection criteria

Inclusion criteria:

• Adults with MSK condition or the at-risk of developing an MSK disorder or condition

• Utilizing wearable sensors

• Sensor placement or assessment of outcomes related to shoulder/upper arm, elbow/forearm/wrist, and hand/finger

• Studies published in peer-reviewed journals and conference abstracts with available full text from 2010

Exclusion criteria:

• Wearable sensors applied on robots

• Exoskeletons

• Brain or human–computer interfaces (BCI/HCI)

• Focusing on gait or balance

• Measurements or applications not related to MSK rehabilitation or assessment

• Inadequate details of hardware or measured outcomes of wearable sensors (insufficient details)

The studies providing details regarding the utilized wearable system on people with UE-MSK condition or at risk of developing a UE-MSK condition were included. Due to variability of sensor placements (and in some cases, due to not providing precise sensor placement locations), sensor placement locations have been divided into three regions: shoulder/upper arm, elbow/forearm/wrist, and hand/finger. Furthermore, due to significant technological advancements of wearable sensors in recent years, only the published studies after 2010 have been included in the search process. General wearable design studies and designs for the assessment of muscle activity or limbs motion (not on a specific population) are also included during the search process. This decision has been made to encompass all possible applications of wearable sensors on MSK-population, especially in motion assessments, a typical evaluation in the clinical rehabilitation field.

Regarding the wearability aspect of the applications, the studies that developed a framework for wearability (e.g., embedding sensors in a fabric pocket) were included. Furthermore, studies that mentioned or considered their presented measurement sensors as a wearable system have also been included.

Exoskeletons, BCI/HCI, and other robotic applications were excluded since they are primarily designed for neurological disorders applications and are outside the scope of this study’s research question. Similarly, the studies focusing on gait or balance were excluded. Furthermore, the wearable sensors recording or measuring general physiological parameters (such as blood pressure and heartbeat) unrelated to UE-MSK conditions were also excluded.

All in all, obtained studies through the database search, a manual search of Google Scholar, and conference abstracts were imported into Covidence, a review management software (https://www.covidence.org). Duplicates were removed, and the remaining titles and abstracts were screened. Subsequently, the full-text review process has been performed to find the eligible studies for inclusion in this review. The references of included studies have been examined to find any potentially relevant articles that were not obtained through the search process of databases.

Data extraction and analysis results

Two independent authors (SM and ES) extracted data on the study designs, study aims, number of participants, sensor placement locations, sensor type and number, outcome(s) of interest, the processing software, and the presence of other features, including home-based applicability, comfortability-assessment, and wireless data transmission ability. We also extracted details on intervention or data collection procedures and the critical points of outcome processing methods of each study.

The studies proposing a novel wearable system for motion assessment that have conducted motion tests on one participant (case-study) or have not conducted any tests on any participants have been categorized as preliminary studies in the study design section.

All the wearable system data corresponding to upper body limbs and segments (shoulder/upper arm, elbow/forearm/wrist, and hand/finger) have been extracted and presented in this review. Several studies might exploit wearable sensors on other body parts such as the neck, back, and knees in addition to upper limbs. Nevertheless, the corresponding information related to neck, back, and lower-body segments has not been reported and summarized. For example, some of the workers’ population studies have used a full-body size wearable system comprising of IMU sensors on lower body parts in addition to upper body regions. Therefore, only corresponding details of hands, wrists, elbows, and shoulder in these certain studies were extracted and analyzed. Subjective outcomes and non-wearable related data are also excluded. Moreover, since the focus of this review is on the applications of wearable sensors, the results and statistical analysis procedures of the included studies have not been included. It must be noted that the full-text script and supplementary materials of all included studies have been reviewed to extract the mentioned parameters. All in all, any potential absence of further details in some of the parameters of studies (e.g., intervention details or software used for processing) is due to the fact that the authors have not reported the corresponding data.

Results

The scoping review search process initially resulted in finding 1644 documents through the 4 databases. Furthermore, 42 studies were also obtained through the Google Scholar. After duplicate removal, 1123 studies were identified for the title and abstract screening stage. Through this process, 976 irrelevant studies were removed, and the full-text reviewing process was initiated. Finally, 67 documents were also removed in this stage, as they did not fit the inclusion criteria or the objective of this review. Hand searching of reference lists led to 11 additional studies at the full-text review phase. This process led to the inclusion and analysis of a total of 80 studies presented in this scoping review. In Fig. 1, the results of searching, screening, eligibility, and inclusion processes have been demonstrated while applying the PRISMA methodology [22].

Fig. 1

PRISMA flowchart of the screening and review process for included studies

The synthesis process led to categorizing the included studies into three categories (1) Clinical population studies, (2) Workers’ population studies conditions, and (3) General wearable design/performance studies. The summarized results for each category are presented in the following. Tables 1, 2, 3 provide the extracted detailed information of included studies in each category.

Table 1 Clinical population studies

Table 2 Workers’ population studies

Table 3 General wearable design/performance studies

Among the identified body of evidence, various researchers from 30 countries have contributed to this field by conducting studies. In Fig. 2, the number of conducted studies in each country has been demonstrated in a bar chart. Researchers of the USA, Canada, Italy, China, and Germany are pioneers in this field by presenting 18, 12, 7, 6, and 6 studies, respectively. It must be noted that in the case of collaborations of researchers from different countries, all author’s affiliations have been considered the origin country of research.

Fig. 2

Number of included studies sorted by the countries

The sensor placement locations were sorted into (1) shoulder and upper arm (scapula and humerus), (2) elbow, forearm (ulna and radius), and wrist and (3) hand, and finger (phalanges). (Fig. 3) Most of the included studies (n = 76) have attempted to measure angles of these limbs (for obtaining Range of Motion or ROM) alongside their angular velocity or acceleration through exploiting IMU, piezoresistive, or optical sensors providing three-dimensional values (represented in Quaternion axes, Eulerian axes, or yaw/pitch/roll). In these studies, subjects performed a series of various joint movements such as flexion/extension, abduction/adduction, and internal/external rotation or a combination of these motions to obtain the measurement parameters of that joint. A summary of the data collection procedures along with the critical points regarding the processing methods used for the translation of obtained joint data into meaningful clinical or non-clinical outcomes have been reported for all three categories in Additional file 2: Appendix B.

Fig. 3

a Sensor placements at different segments for all included studies; b Sensor placements at different segments in clinical population, workers’ population, and general design/performance studies, separately

The outcome column of the following tables summarizing the extracted data of included studies presents the corresponding outcome that has been directly or indirectly obtained via wearable sensor platform. If the targeted outcome of a study is indirectly measured by the wearable sensor platform, the sensor-measured parameters (joint angle, acceleration, or angular velocity) are mentioned in brackets. Moreover, if the outcome obtained by wearable sensors is a secondary outcome of the study, it is demonstrated in that section. Since the assessment and collection of the statistical results of the studies are outside the scope of this scoping review, the related information about the statistical processes and analyses has not been reviewed. We also presented an overall visual representation of the total number of various sensor types used in all the studies, utilizing a pie chart (Fig. 4).

Fig. 4

Number of different types of sensors used in all studies; Examples of others include smartwatches, self-developed systems, LED, and non-IMUs

Clinical population studies

In this category, either the participants of the included study were MSK patients, or the introduced wearable platform was designed and aimed to be applied to individuals with MSK conditions. A total of 31 articles were identified to be included in this category. Nineteen studies out of 31 were related to applications of wearable sensors for shoulder or upper arm conditions. However, the sensor placement of some of these studies included other upper body parts such as the wrist [24,25,26,27,28] and hand [29]. One of the included studies has utilized a wrist wearable system for assessment purposes on patients with distal radius fractures [30]. Three other studies on arthritis and motor impairment patients have used wrist-mounted wearable sensors [31,32,33]. Five studies focused on hand and finger diseases such as hand arthritis [34, 35] and carpal tunnel syndrome [36, 37], applied wearable sensors (such as IMUs, and piezoresistive sensors) that were either embedded in a glove or attached with straps.

The study design types of included articles in this category were case–control studies (12 studies), case series studies (13 studies), and preliminary studies (6 studies), without any Randomized Clinical Trials (RCT) studies (Fig. 5).

Fig. 5

Study design types of included articles in the clinical population studies category

Most of the studies (n = 26, 84% of clinical population studies) exploited IMU sensors in their wearable platform (usually containing a 3-axis accelerometer, 3-axis gyroscope, and 3-axis magnetometer) to obtain joint angles, accelerations, angular velocities, and posture. While only five studies have included EMG sensors in their wearable platform to obtain muscle activity [30, 38,39,40,41]. Some of the studies have also used other instruments and sensors to obtain measurements or treat an MSK condition, such as using high-power LEDs [42] and thermal flexible printed circuit boards for treatment [32] or using piezoresistive or strain sensors for measurements [35,36,37, 43]. In the sensor type and hardware column of Table 1, the model and provider company of sensor, the attachment or embedding method of the sensor in wearable platform, core processing unit, and data sampling frequency of wearable sensors were indicated (in case of reporting by authors).

In the included studies, participants completed functional activities or exercises, like shoulder flexion/extension, abduction/adduction, and internal/external rotation, or wrist, elbow, and fingers flexion/extension for a specific set of repetitions in standardized conditions such as in studies conducted by Duc et al. [38], Kwak et al. [24], and Chen et al. [44]. Some studies have collected the data in a non-standardized set up like a home setting. In these studies, participants’ daily activities have been recorded and analyzed, such as in studies conducted by Pichonnaz et al. [45], Duc et al. [46], Van de Klut et al. [47], Langohr et al. [48], and Haverstock et al. [49]. The explained wearable systems have been identified as home-based wearable systems and have been reported in Table 1. It must be noted that usually, the calibration process of sensors has been described, and a summary of these procedures has been included in Additional file 2: Appendix-B (in case of reporting by authors).

The software utilized for each study’s processing phase (translating raw data into the intended outcomes) has been indicated in the “Software for processing/data display” column of Table 1.Each study’s home-based applicability, comfortability assessment, and wireless data transmission ability have been examined in the final column. In this regard, the introduced wearable system of 17 studies (55% of this category’s studies) has been recognized as home-based applications. Only 1 study (3% of this category’s studies) have assessed the comfortability of their applied wearable sensor platform in subjects [25], and 10 studies (32% of this category’s studies) have explicitly mentioned the ability to send the measured data through a wireless system [25,26,27, 29, 33, 35, 37, 39, 40]. This ability is a crucial parameter for the future development of wearable systems. The share of each sensor placement location in the studies has been demonstrated (Fig. 6).

Fig. 6

Sensor placements at different upper limb segments for Clinical population studies category

Workers’ population studies

In this category, 31 studies have focused on the workers’ population and the risk of work-related musculoskeletal (WMSK) conditions or disorders. Two RCT studies, a crossover study, two case–control studies, 21 case series studies, and 5 preliminary studies are included in this category (Fig. 7).

Fig. 7

Study design types of included articles in the Workers’ population studies category

From the perspective of sensor placement location, 21 Studies (68% of studies in workers’ population category) have placed the sensors on multiple arm segments (the most common segments are the upper arm and shoulder). This can indicate that the most studies of this category have correctly assessed more than one single-point arm segment to obtain a complete perspective of the risk involved in various working conditions. It can be deduced that the primary focus point of these studies is upper arm segment (Fig. 8).

Fig. 8

Sensor placements at different upper limb segments for the Workers’ population studies category

The participants of this category’s studies form diverse samples of different populations including, dentists [55,56,57], surgeons [58, 59], warehouse and manufacturing workers [60,61,62], farmers [63, 64], athletes [65], and other areas. According to Table 2, nearly all the studies have either aimed to investigate ergonomic risk levels of a specific job or measure the exposure to ergonomic risks related to WMSK conditions. A significant number of the studies (about 48% of the included studies) have exploited Rapid Upper Limb Assessment (RULA) score (or a modified version of it), which represents the level of MSD risk for a job task being evaluated [66]. Thus, the RULA score is one of the frequently reported outcomes of a considerable number of included studies of this category. Some of the studies have aimed to evaluate the validity of a proposed wearable sensor platform for monitoring the ergonomic performance of workers with respect to an optical motion capture system accepted as a gold standard [67,68,69].

Regarding the utilized instruments and hardware of wearable sensors applied in the studies, IMU sensors have been exploited in all studies. In this regard, seven studies have declared that Xsens company (headquartered in the Netherlands) products, including MVN link and Biomech™, were used to obtain inertial measurements [55,56,57, 63, 67,68,69]. It can be implicated that due to convenient application, and various provided features, Xsens products are more favorable for researchers aiming to study work-related musculoskeletal risks or non-laboratory setup and locations. Six studies have also utilized sEMG recording electrodes in their system to obtain muscle activity of arm segments [63, 64, 69,70,71,72]. One study has included force sensors in addition to embedded IMU sensors in a glove to acquire complementary information regarding the work-related ergonomic risks of hands and fingers [73].

Two studies have devised a haptic feedback system via a vibration actuation unit in their wearable system design to provide alerts for workers for changing their posture [62, 74]. In these studies, vibrations are applied to the subject’s body, warning the wearer after exceeding a certain threshold of angles and spending time durations in high-risk postures. The applied sensors number, type, model, provider of utilized sensors in included studies are presented in the "Sensor type and number" column of Table 2. The attachment or embedding method of the sensor in the designed wearable platform, core processing unit, the data sampling frequency of wearable sensors, and related data of utilized motion capture systems such as VICON are also described in case of reporting by authors.

Regarding data collection methods or intervention procedures, some of the studies’ participants have performed specific movements in a standardized condition, such as in studies conducted by Humadi et al. [67, 68], Lee et al. [75], and Vignais et al. [76]. Some studies have collected the data in a non-standardized setup like a factory or work setting. In these studies, particular durations of work shifts have been recorded and analyzed, such as in studies conducted by Ohlendorf et al. et al. [55], Blume et al. [56], and Schall et al. [60]. A summary of these procedures’ calibration processes has also been reported in Additional file 2: Appendix-B (in case of reporting by authors).

Regarding the software component of studies, nine articles have stated using MATLAB software for processing or displaying their data and results. Other studies have either not declared the specific used processing software (or platform for developing their software) or have utilized the specialized software presented by provider companies of the sensors. Four studies have mentioned using smartphone apps (self-developed or company-developed) as their data processing software or their provided GUI for subjects [62, 65, 77, 78].

In contrast to the previous section, the home-based applicability is not assessed since the studies are focused on workers’ population. Each study’s comfortability assessment and wireless data transmission ability have been investigated and reported in Table 2. In this regard, 22 studies have stated that their wearable platform has the wireless data transmission ability (71% of included studies in this category), which is a significant finding for the studies of this category. Only three studies have assessed the comfortability of their presented wearable sensor platform through questionnaires and interviews [61, 62, 79].

General wearable design/performance studies

This section explores and summarizes a total number of 18 studies that developed a wearable platform or proposed a novel method or design to assess upper limb movement parameters or muscle activity in individuals without emphasizing a specific context or population (Fig. 9). The wrist, forearm, hand, and fingers are the most common sensor placement locations (n = 8) in this category (Table 3).

Fig. 9

Sensor placements at different segments for the general wearable design/performance studies category

All included studies propose novel designs, platforms, measurement, and processing methods. No patient population or participants from specific contexts are included in these studies. Therefore, the study design types of all included articles are either case-series (n = 12) or preliminary studies (n = 6). The participants of this category’s studies are all healthy individuals (general population). However, in one of the studies, the proposed design is suggested for people who might develop MSK conditions [87], and two studies focused on athlete populations [88, 89].

Another significant point in the included studies of this category is the larger proportion of articles utilizing non-IMU sensors (n = 10.56% of studies in general design category). For instance, piezoresistive, bend sensors, and contact force sensors operating based on sensors’ electrical resistance or conductivity change are more exploited to detect joint angles and motion in the general wearable designs [87, 90,91,92,93]. A novel approach has been considered by using a different type of accelerometer operating based on contact microphones [94]. In this way, the proposed system can detect and obtain accelerations of sensitive motions. Another novel approach has exploited electromagnetic sensing abilities and electromagnetic coils to detect biomechanical motion of joints through electromagnetic impedance [95]. Two studies have utilized sEMG recording electrodes in their general wearable system designs [96, 97]. The related details of wearable sensors, including the utilized number, type, model, and provider of sensors, are included in the “Sensor type and number” column of Table 3. The attachment or embedding method of sensors in the designed wearable platform or the provided wearable platform (like a fabric glove or sleeve), core processing unit and data sampling frequencies of wearable sensors are also mentioned in the case of reporting by authors.

Regarding the software component of studies, the articles have used MATLAB, Python, Microsoft Visual C# software for processing purposes or designing graphical user interfaces. Only one team of researchers has designed a smartphone app to interact with the proposed wearable system [98].

The number of studies indicating home-based functionality or assessment of comfortability of their proposed designs is significantly lower than the two previous sections. Only one study has claimed that their platform has a home-based functional ability [96], and only one study has focused on assessing their system’s comfortability [98]. This can be related to the fact that most of these studies have only considered the general assessment aspect of their platform (n = 16), and these features are of more significant concern in more focused research and areas. Five studies reported their introduced wearable system’s wireless data transmission ability (28% of studies in general design category). The details of each study have been mentioned and illustrated in Table 3.

In the following, some of the significant findings of the result section are summarized and presented. One of the critical aspects is finding the rate of studies aimed to use wearable sensor platforms for either rehabilitation or treatment applications or increasing the quality of remote treatment sessions. In this regard, 11 studies in the clinical population studies category (35% of studies in the category) [26, 27, 29, 32, 33, 36, 37, 40, 42,43,44], three studies in the workers’ population studies category (10% of studies in the category) [62, 80, 85], and two studies in the general wearable studies category (11% of studies in the category) have focused on treatment applications [87, 98].

Other investigated features in the synthesized studies were the home-based applicability and the presence of any sort of comfortability assessment. Home-based applicability feature has been explored in the first and third categories (Clinical population studies and General wearable design/performance studies); however, the assessment of comfortability has been explored in all three categories. In this regard, 18 studies in total have reported a home-based assessment or treatment (37% of included studies in clinical population and general design categories). The common element in these studies is preparation and utilization of a convenient wearable platform for subjects in the form of a glove [35]; a band, strap, or watch [25,26,27,28, 31, 33, 43, 44, 98], a sleeve [29], and a shirt [47,48,49, 53]. On the other hand, only five studies (7%) have assessed the comfortability of their wearable systems in subjects through methods such as questionnaires [25, 79] and interviews [62, 79].

Another significant finding is the high and low rate of using IMU (n = 68, 85%) and sEMG (n = 13, 16%) sensors in the included studies, respectively. In total, 13 studies have exploited wearable sEMG systems to obtain muscular activity of subjects. Some studies have used both IMU and sEMG sensors to obtain both joint motion and muscle strength information of participants [30, 38,39,40,41, 63, 64, 69,70,71,72]. Eight studies have placed sEMG electrodes on upper arms and shoulders, four studies have located sensors on forearms, and one study has recorded both upper arm and forearm activities.

The sensor placement locations of each category were presented in previous sections. In Fig. 3, all categories’ sensor placement locations and the number of utilized upper limb segments in wearable sensor systems of studies have been demonstrated to compare each category (Shoulder/Upper arm, Elbow/Forearm/Forearm/Forearm/Wrist, and Hand/Finger). These systems may place the sensors on multiple segments of the upper limb [28, 67, 96]. Thus, all the segments included in each study have been counted separately in corresponding figures (for instance, if the sensors were placed on the forearm and upper arm, both segments were included in the corresponding figures). In total, 129 sensor systems have been utilized in the included studies.

Designing or using a smartphone application for monitoring or displaying the obtained data in several studies was another significant finding of this review. In this aspect, ten studies have either used or developed a smartphone app in all three categories [26, 27, 31, 33, 62, 65, 77, 78, 89, 98]. Considering the remote data collection method in some studies, this ability can facilitate the supervision and connection of researchers, clinicians, and patients.

Discussion

This scoping review summarized 80 research studies that addressed and found a variety of wearable sensors applications in assessment, prevention, and treatment of upper extremity MSK conditions with the most common goals of joint motion measurement. Reviewed papers were sorted into three categories of clinical population, workers’ population, and general design/performance studies as three primary application fields. The populations of clinical studies were patients with UE-MSK disorders, while the participants of workers’ population studies were manufacturing workers, athletes, surgeons, farmers, and healthy individuals at risk of developing an MSK condition at work. Prevalence rate of MSK or UE-MSK conditions among US workers’ population was 8.23% according to a conducted study in 2018 [105]. Moreover, 1.7 billion people are dealing with MSK conditions at global scale [17]. Thus, the reviewed information of wearable systems in each category can provide a separate overview of different populations categories and address the available knowledge gaps of each. The third category of General wearable design/performance studies proposed novel systems or settings that were not attributed to a specific patient population or context. This category proves to be advantageous for researchers or designers to be informed about available wearable systems design ideas for their future studies regardless of any specific population or setting.

Clinical population, workers’ population, and general wearable design studies included 25, 24, and 12 observational studies (such as case series, and case control studies), respectively. There were 17 preliminary studies (21%) proposing a novel design, or a processing/classification method were also found and reviewed in all included studies (Fig. 10). A considerable number of included studies in all three categories are conference abstracts that generally represent initial findings on new applications. It may indicate that these applications did not always progress to fully powered studies although it may also reflect the stage of innovation in this field. Two RCT studies were included in the workers’ population category, while no RCT studies were found in the clinical population studies category. Conducting RCTs will obtain more valid results regarding the validity of measured outcomes by wearable systems in the field of UE-MSK disorders. All in all, it appears that studies with more rigorous study designs providing more valid data are still not conducted or scripted. This finding is in accordance with results of [106] and [107], in which the number of clinical trials is either low (eight clinical trials including the use of wearable sensors systems in [106]) or reported to be lower than studies on healthy populations [107]. This gap can be addressed by introducing novel and easy-to-apply wearable systems that can facilitate conduction of various tests on a sufficiently large sample MSK patients and healthy individuals. The collaboration of rehabilitation engineers and clinicians in a study on the applicability of wearable sensors in UE-MSK can also cover both technological and clinical aspects. This association is recommended as a solution to overcome this challenge, as already suggested by Collinger et al. to address similar challenges found in the neural interfaces and integrated prosthetics field [108].

Fig. 10

Study design types of included articles in both clinical and workers’ population categories

One of the challenging points of included papers is the wearability aspect of utilized sensors and instruments. The included studies have either developed a wearable framework for sensors or have called their system a wearable design. However, according to the perspective of other researchers, "wearable sensors" are defined as electronic systems and computers integrated into comfortable and wearable clothing and other accessories [109]. In a rather distinct definition, implanting sensors in the body can also be considered as developing a wearable system [110]. A considerable number of included studies have merely attached IMU, sEMG, and other sensors to the subjects’ bodies through adhesive tapes and other materials and did not mount, embed, or integrate the sensors into a fabric, textile, or structure. Although they have called and considered their system a wearable design platform, the convenient wearability feature of their system is uncertain. This issue can be attributed to the lack of a widely accepted general definition of wearable sensors. Providing such definitions and standards can appropriately address this issue. In the case of this review, the mentioned studies have not been excluded since the aim of a scoping review is to map all available evidence of the wearable sensors field [111]. Therefore, all these studies have been included to provide a better overview of research contributions.

Another significant point is the low rate of treatment applications in the included studies. In total, only 20% of all studies (n = 16) have considered a treatment application in their reports. These studies have used wearable sensors for unsupervised or indirect treatment. In workers’ population category, 3 studies developed a feedback method that prevents the subjects from developing an MSK condition during working hours. The low rate of treatment applications can be attributed to the fact that clinicians or researchers usually face difficulties in providing patients with wearable sensors that can record and store information for long durations. Another potential cause can be related to the complexity of working with wearable sensors. Investigating the rehabilitation and treatment applications in various conditions and their potential barriers is strongly recommended for future studies. Other studies (n = 64) focusing on assessment applications of wearable sensors provide motion information. Researchers of these studies have aimed to gain valuable data about the treatment process or work risk factors of MSK conditions. The greater number of assessment studies can be related to technological advancements of IMU sensors and big data field [112]. A high rate of wireless data transmission can store and send more than 50 samples of joint orientation and motion per second according to the reviewed papers of this scoping review. Nonetheless, the low rate of using a feedback procedure (visual, auditory, haptic, and other feedback forms) with the available wearable systems can be one of the reasons of low rate of wearable sensors treatment applications in MSK conditions.

Another encountered challenge of this review was finding similar articles about one identical study published by similar research teams. These articles were mainly reported as a single study in this review, and it has been attempted to summarize all essential information of these articles. However, in several cases such as Duc et al./Pichonnaz et al. [38, 45, 46], Burns et al. [26, 27], Humadi et al. [67, 68], and Vignais et al. [76, 79], all found articles have been reported separately since either the interventions and processing methods, wearable instrument, or reported outcomes differed from each other. Therefore, they have been treated as separate studies.

One of the investigated features in the studies was their home-based applicability, and as mentioned in the results section, 37% of studies (n = 18/49) in the clinical population and general wearable design categories assessed or suggested a home-based application of their wearable system. This rate is in accordance with reports of other wearable sensor studies, in which the ability of long motion recording activity or tracking daily activities of participants has been considered challenging [113, 114]. Therefore, enabling home-based motion recording of individuals for long durations deems necessary for future research applications considering that the subjects must be able to set up the settings conveniently and independently at their home for enabling the opportunity of conducting more robust studies. Moreover, the comfortability parameter of the applied systems is another critical aspect that has been investigated in a low share of studies (n = 5, 7%). This feature has a significant impact on the extensive functionality of wearable sensors, especially for recording long daily activity recordings. This rate represents another gap of knowledge in this field. Conducting studies that can provide reliable and valid information on the comfortability of wearable sensors through standard questionnaires and interviews or potential challenges that emerged in using these kinds of systems can guide the developer companies of these systems toward more generalizable wearable systems. Therefore, it is highly recommended that researchers consider the comfortability assessment of their applied wearable sensors in subjects for their future studies.

Most of the papers (n = 68, 85%) have used IMU sensors to obtain inertial data of subjects. However, sEMG recordings were also made in some of the studies (n = 13, 16%). According to these results, ROM and joint angles, or the time spent in certain angle ranges were more considered and measured as the primary outcomes of studies. While muscle strength and maximum voluntary muscle contraction were usually measured as complimentary outcomes. This finding can be primarily attributed to the fact that the sEMG costs, setup process, finding precise sensor placement location, and providing settings for extended recording ability (continuous skin cleaning process) are troublesome procedures [115,116,117]. While using IMUs do not necessarily require advanced settings or setup [115]. They can operate in wireless setting and without causing notable disturbance in comparison to sEMG sensors. In this scoping review, only two of the studies have proposed a convenient wearable platform for positioning and utilizing sEMG signals [96, 97]. Other studies merely attached the electrodes with straps or adhesive tapes. As mentioned by the authors of this study, this leads to the emergence of challenges in finding accurate and sEMG sensor placement and long extended signal recording durations for future measurements. Thus, providing a comfortable and easy-to-use wearable platform for sEMG electrodes can ameliorate the utilization rate of sEMG systems in future studies.

Regarding sensor placement locations, no significant difference is noted between upper arm/shoulder and forearm/elbow/wrist regions. However, the number of studies that have focused on hand/fingers region is considerably lower than other regions in clinical population and workers’ population studies. It can be speculated that this is due to the convenient placement setting of IMU sensors on upper arm and forearm regions.

Another notable point is the similarity of processing methods of the included studies. The process of obtaining outcome in all three categories of studies usually contain a filtering procedure to clear the noise from obtained data and a classification step to detect a joint movement. Low and high pass filters perform inertial data filtering tasks that result in a bandpass filter. The frequency range of this filter is not unique in the studies; however, the cutoff frequencies are between 0.1 Hz [28] for lower cutoff frequency to 10 Hz [101] for upper cutoff frequency. This finding is consistent with similar reported filter frequencies by Kim et al. [19]. Moreover, the EMG signal filtering is also performed by a bandpass filter. The reported frequencies are between 10 Hz (for lower cutoff frequency) to 500 Hz (for upper cutoff frequency) [30]. On the other hand, the classification step is realized through various techniques. One of the methods to perform classifications is machine learning algorithms like Support Vector Machine (SVM), Convolutional neural network, and k-nearest neighbors, that are applied in studies such as Burns et al. [26], Jang et al. [99], Nath et al. [84], and Rodríguez-Vega et al. [73]. Another classification method is simple thresholding, which is considered by Pichonnaz et al. [45], Duc et al. [38], and Larrivée et al. [25]. All these classification methods have unique advantages and disadvantages that must be utilized according to the study’s aims and applications. Further investigation of these methods is beyond the scope of this paper. However, important points about each study’s applied processing and classification methods have been reported in Additional file 2: Appendix B of this review. As mentioned in previous sections, the statistical analysis of studies has not been reviewed or summarized in this review. Thus, the utilized software for statistical analysis purposes of studies has not been reported. Nonetheless, the related analytical procedures have been usually conducted on different versions of MATLAB, Excel, and SPSS software.

Exploiting smartphone applications for monitoring or displaying the obtained data was another investigated feature in the studies of this review. To provide a better experience for subjects or patients, especially for work environments in which subjects might not have easy access to computers or displays, developing or using a smartphone app can significantly improve the quality of monitoring or displaying the inertial data [118]. Moreover, through smartphone applications, various forms of feedback such as auditory, text, and vibrations can be provided for users to notify them regarding their risky postures or correct/incorrect motions [118]. Considering the significant technological advancements in smartphones, developing a smartphone application can raise the quality of supervision and connection of health care providers, clinicians, and researchers with individuals. Therefore, including a smartphone application in their wearable system is highly recommended for future applications considering its current low rate of utilization in UE-MSK conditions (n = 10, 13%).

As can be understood from the results of this scoping review, a diverse set of equipment, setting, processing and intervention has been applied in studies that used wearable sensor systems. No globally accepted standard method for sensor placement, hardware characteristics, intervention guidelines for wearable systems, extracting features from the inertial signal, or categorizing activity across important functional motions has been proposed. In accordance with findings of Dobkin and Martinez [112], and Attal et al., [119], this can be considered as one of the most important limitations of this field.

The findings of this scoping review are limited to the included studies and synthesized pieces of found evidence, and this is one of the limitations of this review. Conducting a broad search with other keywords (for example, including different types of MSK disorders like arthritis and carpal tunnel syndrome in the search process keywords) and database might result in finding other studies that alter the interpretation of this review’s results. Nevertheless, in this scoping review, it has been attempted to include all necessary keywords to yield all related papers. While other search strategies might lead to different results. Other available sources have been also searched and examined to acquire all available pieces of evidence in the field. One of the other points that should be noted when reading our scoping review is that we did not consider the statistical analysis/results, including details about algorithms and performance metrics. However, we would like to clarify that the primary focus of this scoping review is on wearable sensor systems and their corresponding hardware. Given the breadth and depth of the topic, we made a deliberate decision to exclude detailed statistical analyses and results to keep the manuscript concise and in alignment with our specific research scope. We believe this approach allows us to provide a clear and coherent overview of the wearable sensor technologies and their applications.

Conclusion

Wearable sensors are increasingly applied in UE-MSK related studies. They prove to be significantly important in the development of the next generation of health care technologies in the assessment and treatment fields. They can be utilized for clinical assessments, home-based applications, daily activity recording of patients and individuals or non-standardized areas such as work environments to characterize the physical risk factors of developing UE-MSK conditions or investigating the ergonomic risk levels of various work environments. A large share of reviewed papers are observational studies that shows the need for conducting more robust research studies from the aspect of study design to yield more valid results. IMU sensors were primarily applied in most of studies (n = 68), and sEMG sensors were included in wearable platform system of some papers to obtain inertial data and muscle activity (n = 13), respectively. The share of assessment-oriented studies is greater than treatment-oriented papers that represents the current assigned primary role of wearable sensors in obtaining objective upper body motion data. Based on this review’s findings and current evidence, conducting randomized clinical trials using wearable sensors system and research about novel treatment applications of wearable sensors, focusing on home-based applicability of suggested systems, the inclusion of a variety of sensors such as user-friendly EMG and bend sensors in addition to IMUs, and designing smartphone applications for convenient and continuous monitoring of users are recommended for future studies of this field.