The Use of Wearable Sensor Technology to Detect Shock Impacts in Sports and Occupational Settings: A Scoping Review

Shock impacts during activity may cause damage to the joints, muscles, bones, or inner organs. To define thresholds for tolerable impacts, there is a need for methods that can accurately monitor shock impacts in real-life settings. Therefore, the main aim of this scoping review was to present an overview of existing methods for assessments of shock impacts using wearable sensor technology within two domains: sports and occupational settings. Online databases were used to identify papers published in 2010–2020, from which we selected 34 papers that used wearable sensor technology to measure shock impacts. No studies were found on occupational settings. For the sports domain, accelerometry was the dominant type of wearable sensor technology utilized, interpreting peak acceleration as a proxy for impact. Of the included studies, 28 assessed foot strike in running, head impacts in invasion and team sports, or different forms of jump landings or plyometric movements. The included studies revealed a lack of consensus regarding sensor placement and interpretation of the results. Furthermore, the identified high proportion of validation studies support previous concerns that wearable sensors at present are inadequate as a stand-alone method for valid and accurate data on shock impacts in the field.


Introduction
Every load applied to the body will lead to an impact force on the joints and the articular cartilage, to which the body must respond with shock attenuation. This is not necessarily harmful-it is well known that loading is a prerequisite for good joint health, as mechanical forces contribute to maintain the specialized function of articular cartilage and subchondral bone [1]. However, at the same time, excessive strain, either over time or as a result of a single impact with abnormal stress, may cause damage to the joint, which in the short term can lead to pain and reduced function and in the long term to the development of osteoarthritis [2][3][4]. Furthermore, the body can be exposed to shock impacts that may hurt other structures-muscles, bones or inner organs, and of particular concern, the brain. To identify and define the threshold for shock impacts that may cause damage will be of utmost relevance to define tolerable training and activity loads and to prevent both acute and stress-related injuries. However, to explore this, we first need to gain a better and more in-depth understanding of which methods can monitor shock impacts most accurately in real-life settings in different contexts.
Wearable sensor technology provides an alternative to classical laboratory-based assessments of human performance that enables real-time monitoring in natural environments [5], without the cumbersome set-up procedure and limitations related to space [6]. Assessments of human motion by wearable sensor technology generally utilize Inertial Measurement Units (IMUs), which usually consist of triaxial accelerometers, often complemented by three axis gyroscopes and/or three axis magnetometers [7]. The rapid growth in commercially available wearable sensors makes them easily accessible and affordable [8].
In this study, we will use a scoping review methodology to map the use of wearable sensor technology to detect shock impacts in two domains: sports and occupational settings.

Sports
Wearable sensor technology in sports has been reported to address injury prevention, motion analysis, technique classification, and performance and skill-level assessments [9]. Existing studies have addressed a wide audience, spanning from youth sports and recreational activity to top elite athletes [8][9][10]. In the past years, numerous studies have also included IMUs in combination with global positioning systems (GPS) and heart-rate monitoring to investigate various training paradigms in different sports and assess the balance between improvements in athletic performance and risk of overtraining and potential injuries [11,12]. Training load or stress is within this context typically defined as the sum of external (training volume and intensity) and internal (physiological parameters and perceived exertion) loads combined [12][13][14]. Many of these studies have used wearable sensors to monitor the total exposure over time-during a training session, a period of training or a whole season, most often in team sports [15][16][17][18], and in running [14,19,20]. However, within the sports domain, numerous types of sports and activities involve shock impacts that may be harmful from just one or a limited number of impacts-alpine skiing, running, gymnastics, team sports such as soccer, volleyball, and different types of invasion sports such as rugby and football, to mention a few. In this study, we will focus specifically on the use of wearable sensors for detecting shock impacts at these events.

Occupational Settings
Wearable sensor technology enables assessments of real-time measurements of workers in their natural setting. Spook et al. [21] recently investigated worker needs and preferences with regard to implementation of sensor technology to measure and monitor physical job demands and work exposures, and they found that workers were positive to such methods and could see the value of both real-time feedback and access to data on demand. However, a recent narrative review from Lim and D'Souza [22] states that despite growing interest, research utilizing wearable sensor technology in applied occupational settings is still sparse. Their findings reveal that the use of inertial sensors for biomechanical exposure assessment in occupational settings has gained increased attention the past few years, addressing different contexts including healthcare facilities, shift workers in different physically demanding jobs, as well as white-collar work. However, it may seem as though quantification and descriptions of postural information have been the primary scope in existing studies. Awkward postures are regarded as a determinant for the development of musculoskeletal disorders and is therefore an important measure to quantify in occupational settings. Still, depending on the occupational group in question, objective measurements of other exposures may also be important. As in the sports domain, some workers are repeatedly exposed to shock impacts. Construction workers using tools with high single accelerations, such as nutrunners, is one example. The operation of high-speed boats is another. Reduced physical performance following a large number of high mechanical shock impacts during a high-speed boat transit has been shown [23]. The suggested mechanisms behind the degradation on performance were muscle damage and localized muscle fatigue due to repeated eccentric contractions to attenuate the shock impacts. Furthermore, high-speed boat crew members often experience acute traumas and chronic musculoskeletal pain related to the nautical activity [24][25][26]. In some instances, shock impacts can be measured on the objects, as in the high-speed boats. However, as shock waves are attenuated by biological tissues within the body, the effect of the shock impacts on different anatomical structures can be very different from that measured on the boat itself. Therefore, in this study, we are looking into whether there are existing methods to objectively measure shock impacts using wearable sensors in occupational settings.

Study Aim
It appears that research utilizing wearable sensors both within the sports domain and in occupational settings primarily is recognizing the value of monitoring total exposure, risk factors related to activity, overloading or ergonomics, and resultant consequences (such as fatigue or injuries), rather than the specific events that may cause harm. Furthermore, it seems as though the utilization of wearable sensors for specific events has come somewhat further in the sports domain. There is a need to examine whether this impression is true, or if current literature also presents reports on the use of wearable sensor technology to detect undesirable external stimuli in the form of potentially harmful shock impacts in sports and occupational settings. To come further in defining thresholds for whether joint loads are beneficial or damaging in different contexts, objective data from real-life settings quantifying single shock impacts are warranted as a supplement to existing methods that monitor total load over time or exposures to ergonomic strain. Therefore, the main aim of this scoping review is to present an overview of existing methods for assessments of shock impacts using wearable sensor technology within sports and occupational settings. The following research questions will be addressed:

1.
What type of wearable sensor technology is used to measure shock impacts? 2.
In what types of activities is wearable sensor technology used to measure shock impacts? 3.
Which sensor placements and outcome measures are used when measuring shock impacts using wearable sensor technology? 4.
Which knowledge gaps are apparent in the literature regarding wearable sensor technology for assessments of shock impacts within sports and occupational settings?

Materials and Methods
Scoping reviews serve to synthesize evidence and assess the scope of literature on a given topic. In contrast to a systematic review, a scoping review does not aim to synthesize evidence from the included papers in order to evaluate study quality or provide evidence to inform practice. Rather, the main purpose is to map the available evidence.

Protocol
This scoping review follows the scoping review methodological framework to systematically map a research area, identify the main sources and types of available evidence, and identify research gaps in the existing literature [27]. The scoping review protocol was drafted using the PRISMA methodology and its extensions for scoping reviews (PRISMA-ScR) [28]. Results are reported in line with Moher et al. [29].
The authors responsible for this study compose a cross-scientific team with expertise in physiology, sports medicine, biomechanics, movement analysis and wearable sensor technology. The study protocol was established from discussions on the overall aim, selected fields to be included, and the context and limitations of the research questions to be addressed. From this, the team agreed on search terms, inclusion and exclusion criteria, search strategy, and selection of databases to search.

Eligibility Criteria
To be included in the data material for this scoping review, papers had to measure shock impacts on the whole body or specific body regions with the use of wearable sensor technology. Wearable sensors were limited to either sensors attached to the body, to clothing, or to equipment used by the individual. Furthermore, the measurement methods needed to be relevant in a sports or occupational setting. Full-text review articles were retrieved if they included relevant measurement methods of shock impacts, but they were not included in the data material. Peer-reviewed journal papers or conference papers were included if they accommodated the initial criteria, were written in English, and were published between 2010 and 2020. Due to the rapid development of this field, a 10-year limitation was chosen to narrow the results to relevant sensor technologies likely still in use, with the rationale that technology developed before 2010 still in use would appear in publications within the chosen time frame. Papers were excluded if they did not fit the purpose of the study, the included subjects were patients or individuals with a functional impairment, disability, or illness, or if they focused on movement quantification, total load, movement quality, or technique only. Regular gait analysis was deemed not relevant for the purpose of this paper. Papers on gait or walking were eligible for inclusion only if related to a specific work-related context and if they specifically reported measurement of impact forces.

Information Sources
To identify possibly relevant publications, on 7 September 2020, the following bibliographic databases were searched: MEDLINE, SportDiscus, Scopus, PubMed, Compendex, and ISI Web of Science. The search strategies were drafted by the help of a librarian (Trude Eikebrokk) and further refined through discussions between the three authors. The final search strategy for MEDLINE can be found in Appendix A, Figure A1. The final search results were exported into EndNote and then into the software tool Covidence [30]. Duplicates were removed in Covidence.

Selection of Sources of Evidence
To ensure consistency among reviewers, all three authors performed a title and abstract screening calibration for a random selection of 25 papers. The results were compared and led to minor adjustments of the screening tool. Then, the process was repeated with another 25 randomly selected papers before the screening tool was amended once more to assure consistency between reviewers before the screening of all abstracts started. The final screening tool for the title and abstract screening and the full text screening can be found in Appendix B, Tables A1 and A2. The software tool Covidence was used for extraction of relevant titles and abstracts, and then for the extraction of eligible papers. First, the title and abstracts, and thereafter the full text publications of all included abstracts, were screened by two reviewers to identify papers for final inclusion in the data material. Disagreements on study selection were resolved by discussion among all three reviewers, first when screening titles and abstracts, and later when reading all eligible full-text papers.

Data Charting Process
A data-charting form was developed by the three reviewers to determine which variables to extract from the eligible studies. One reviewer was in charge of charting the data and consulted the other two whenever there was any doubt as to what data should be extracted. Then, the results were discussed.

Data Items and Synthesis of Results
The type of data extracted from the eligible studies were article characteristics (e.g., country of origin, study design, aim or purpose), article methods (main outcome measurement of relevance to this review, population used, type of sports, impact site), types of wearable sensor technology used, additional technology used for validation, and the key findings of relevance. We grouped the studies by types of activities investigated. The data chart can be found in Table 1.          The country of origin, study design, participants, type of wearable sensor technology used, sensor placement, and additional technology for validation were extracted and counted ( Table 2).

Additional Technology for Validation *
Force plate/instrumented treadmill (15), camera/video recordings (14), GPS enabled watch (1), handheld dynamometer (1), Vertec Jump Trainer (1), photo-electric timing system (1) * Several studies have split origin between two countries and have utilized more than one type of wearable sensor and/or additional technology; thus, numbers may sum up to <34.

Results
The initial search resulted in 628 abstracts that were screened by two independent reviewers. Of these, 543 were not accommodating the inclusion criteria. Eighty-five publications were selected for full-text screening. Following the reading of these papers, 51 additional papers were excluded, leading to a final number of 34 included papers in the data material ( Figure 1). The majority of the included papers presented wearable sensor measurements of landing impacts during foot strike in running (number of studies included = 12), head impacts in invasion and team sports (soccer, football, rugby, and lacrosse) (number of studies included = 10), and impacts in different jump landings on feet, including somersaults (number of studies included = 6). One of the studies classified as measuring impact No papers accommodating the study criteria for final inclusion reporting measurement methods for shock impacts in occupational settings were identified. To assess whether the search results in the occupational domain were due to a too narrow search strategy, we performed a new search on 21 October 2020 in the same databases and included "military", "personnel", "police", "speed boat*", and "construction" in the search string. This did not result in any additional eligible studies. Thus, the final extraction includes papers in the sports domain only. Of the 34 full-text papers included from the sports domain, 33 were peer-reviewed journal articles , and one was a conference abstract [64]. The included papers are listed in Table 1 with the following information included: author(s), year of publication, type of sport, study design, study aims, main outcome measure, population, type(s) of wearable sensor technology and placement of sensor, additional technology for validation, and key findings. With regard to the main outcome and key findings, the information extracted in Table 1 is limited to the outcome of relevance and key findings of relevance for the purpose of this scoping review and the defined research questions. An excerpt of the demographic and technological characteristics of the included studies is given in Table 2.
The majority of the included papers presented wearable sensor measurements of landing impacts during foot strike in running (number of studies included = 12), head impacts in invasion and team sports (soccer, football, rugby, and lacrosse) (number of studies included = 10), and impacts in different jump landings on feet, including somersaults (number of studies included = 6). One of the studies classified as measuring impact during jump landings on feet also included landing on hands in gymnastics. The remaining included papers presented the following other main outcomes: breaking and stopping in running (n = 1), whole body impact in rugby (n = 1), landing, cutting, and stepping in badminton (n = 1), GRF predicted from trunk-mounted sensor in running (n = 1), change in mechanics during running derived from peak COM acceleration (n = 1), and foot strike during multiple motions/plyometrics (n = 1).

Discussion
The main aim of this scoping review was to present an overview of existent methods using wearable sensor technology for assessments of shock impacts within sports and occupational settings. We specified four research questions:

1.
What type of wearable sensor technology is used to measure shock impacts? 2.
In what types of activities is wearable sensor technology used to measure shock impacts? 3.
Which sensor placements and outcome measures are used when measuring shock impacts using wearable sensor technology? 4.
Which knowledge gaps are apparent in the literature regarding wearable sensor technology for assessments of shock impacts within sports and occupational settings?
Surprisingly, no papers accommodating the criteria for final inclusion were identified in the occupational domain. Thus, the lack of studies assessing shock impacts using wearable sensor technology within occupational settings may be stated to be the overall biggest knowledge gap identified in this study.
High-speed boats personnel and some types of construction workers are examples of workers known to be exposed to repeated shock impacts, which are believed to negatively affect health and performance. For industry workers, exemplified from the petroleum industry, walking on hard surfaces and climbing stairs for long shifts, five days a week, year round wearing heavy and stiff safety shoes is believed to cause musculoskeletal disorders in the lower back and lower limbs [65]. In work environments such as these, it is the high repetitions, the long durations, and also the accumulated load of different unwanted exposures that pose a risk [66]. As stated previously, quantification and descriptions of postural information has been the primary scope in existing studies within the occupational domain. However, if one wants to establish associations between exposures and health and performance outcomes, it is important to include the measurement of several hazardous exposures to account for the accumulated occupational load. On some occasions, relevant measures will include the dose and frequency of shock impacts during work. Thus, further research into the assessments of shock impacts using wearable sensor technology within occupational settings seems warranted.
This scoping review indicates that the urge to find technology for assessing shock impact has been higher in the sports domain than in the occupational health area. The use of wearable technology in sports is driven by the need to understand biomechanics to prevent injury and to provide immediate feedback to athletes and coaches, which is a common interest for both. Although the benefits of wearable sensors systems (e.g., shock impact assessment) have significant potential for health risk mitigation in the occupational health domain, their ability to capture specific information that may be considered personal and private by the worker may hinder its utilization [67]. Hence, the potential conflict of interest by employers and workers may possibly explain the lack of research in this area as well as the lack of papers on shock impacts in the occupational domain. Another argument is that occupational health assessment is driven by standards and methods for risk assessment to prevent accidents from happening, that employers and workers do not see the usefulness of these functions in enhancing workers' safety and health. Furthermore, the use of new methods for assessment has a much longer timeline for implementation and is still immature for many occupational health workers and their management.
Due to the absence of studies on occupational settings, the remaining part of this discussion will be concentrated on addressing each of the four research questions for the sports domain.

What Type of Wearable Sensor Technology Is Used to Measure Shock Impacts in Sports?
All the included studies in this scoping review included some kind of accelerometer, except for four studies that used insoles with force sensors. The type of accelerometer varies, from uni-to triaxial, and it also varies whether the accelerometers were embedded in an IMU that also included a gyroscope and a magnetometer. All studies had a description of the type of sensors used; however, the level of detail was somewhat inconsistent, making it challenging to compare differences among the different types of accelerometers used. More than 20 different types or brands of sensors were utilized in the studies, but few studies had any line of argument on why they had chosen this particular brand or type. The majority of the studies provided a description of the number of sensors used and sensor placement, allowing others to replicate the study protocol. However, the level of detail varied also on this aspect.
With regard to which type of data most often is extracted from the accelerometer/IMUs, peak acceleration is the dominant outcome measure. This applies to studies assessing running (foot strike), different landings, as well as head impacts. For studies utilizing insoles, force and pressure distribution is the outcome of interest. A detailed discussion on the outcome measures considered in the different contexts will be provided later in this discussion.
Of the 34 studies included, as many as 25 utilized additional technologies to the wearable sensors. For studies on running, seven studies used force plates or instrumented treadmills with force plates embedded [34,37,44,47,49,57,63], two used force plates and camera recordings [55,60] and three used camera recordings only [38,52,59]. For assessments of landing impacts, force plate data was reported in four studies [31,33,62,64], force plate and camera recordings were reported in one study [32] and camera recordings only were reported in one study [46]. The main reason for adding these measurement technologies was that the studies had a single intervention design, with validation of the data collected from the wearable sensor technologies as part of their main purpose. For studies concerning head impacts in invasion and team sports, camera recordings were added in seven of the included studies [36,40,41,48,53,54,58]. The combined use of wearable technology and other technologies, in many cases representing current gold standards, will be discussed in detail in the last section of this paper where identified knowledge gaps are addressed.

In What Types of Activities Is Wearable Sensor Technology Used to Measure Shock Impacts in Sports?
Two areas were identified to stand out regarding this question: peak impact during foot strike in running and head impacts in invasion and team sports such as football, lacrosse, rugby, and soccer. Thereafter, landings after jumps or plyometric activities seem to be given some attention in activities such as gymnastics, ballet, and non-invasive ball sports such as badminton.
Outdoor team sports is one of the areas were GPS early on was utilized as a method for monitoring player movements [68]. In addition, the use of Micro-Electro-Mechanical Systems (MEMS) technology including different types of IMUs and magnetic sensors in this area has expanded, allowing further assessments of total workload, technique, and physical performance [6,69]. However, it became apparent from this review that single shock impacts-other than head impacts-are not presented in the literature, neither for contact team sports such as soccer, football, or team handball nor other team sports such as volleyball and basketball. This may be somewhat surprising, given that several of these sports are characterized by jump landings, rapid changes of direction during high velocities, running, and pivoting activities, which all cause considerable loads on the lower extremities. It is further well known that athletes in these sports are exposed to considerable risk of injury to the lower extremities [70][71][72][73], both acute injuries such as meniscus and ligament tears, cartilage lesions, and fractures, and overuse injuries such as degenerative meniscal tears and jumper's knee, to mention a few. For the latter case of overuse injuries, it can well be argued that monitoring total load and/or exposure over time may be the most relevant measure. This is also the impression from the current research literature. Even though the search strategy for this scoping review did not include "total load" or "movement quantification" as terms, we still identified and excluded 21 studies that had total load or movement quantification as their primary outcome, and numerous more studies on this topic can be found. Studies assessing total load will typically emphasize the number of jumps, jump height, total distance covered in running, velocity, and/or changes of direction during a game, a season, or a defined period of training and competition [68,74]. Differences in total load exposure between athletes with different positions on the field are also commonly reported [75][76][77], and the literature further reflect the inclusion of athletes at different ages and levels of performance. However, even though such data can be valuable for monitoring exposure, helping balance total training load, and developing strategies to prevent injuries, it could still be suggested that more insight into the effect of each shock impact event could be of interest. For example, the joint load exposure of a total number of jumps performed during a volleyball season will be strongly dependent on the forces that need to be absorbed during landing. Jump height can give an indication about the forces absorbed during landing. However, to identify single situations that imply such high loads that they present particular risk for either an acute injury or a rapidly developing overuse injury, specific assessments of shock impacts are required. An interesting observation in addition to the identified lack of studies on shock impacts in team sports including jumping is that no studies accommodating the inclusion criteria were found for track and field. This is particularly noteworthy for the high jump, the long jump, and the triple jump. These events imply high impact forces, especially in the last one-legged plant step before take-off. Measurements on the magnitude of the single shock impacts during this critical phase of the jump would probably be interesting both for assessments of performance and for injury prevention/load management in training.
In contrast to team sports, it is interesting to note that the approach seems different in studies on running. Even though total exposure (distance covered, surface, shoes) and running technique (forefoot, midfoot, or rearfoot landings, stride length) are topics under investigation in many studies on running, attention is also explicitly directed on single landing impacts: 15 studies on running [34,35,37,38,43,44,47,49,50,52,55,57,59,60,63] accommodating the inclusion criteria were identified. From these studies, knowledge on foot strike impacts can partly be transferred to several of the mentioned team sports.
However, this will only be valuable for quite straightforward running-not changes of direction, breaking, stopping and pivoting, which are the movements causing the most strain on the joints and thus have the highest potential for causing damage.
Only one study was identified and included from winter sports: assessing drop landings in snowboard [64]. However, this was a conference abstract with information missing on design, primary outcome, and participants. The almost complete lack of studies investigating shock impacts during demanding winter sports came as a surprise. Several winter sports are conducted under harsh conditions, with considerable variation in the external demands posed by the environment [78]. As with contact team sports, the injury risk is high in winter sports such as alpine skiing, freeskiing, and snowboard [19,79,80]. Research on these sports relevant for sport-specific demands and injury prevention has been done under controlled laboratory settings. However, this does not necessarily transfer to real-life settings. The use of wearables in field studies in these sports seem to be concentrated on performance assessment and enhancement with evaluation of technique (for example turn switch detection) and tactical dispositions [78,81,82]. Another winter sport with considerable landing impacts is ski jumping. No studies on shock impacts during landings in ski jumping were identified for inclusion in this review. However, there are studies combining IMUs with force insoles, with attention toward the positioning of the skis and the landing technique [83,84]. The lack of studies investigating shock impacts during real-life performance in demanding winter sports is identified as a knowledge gap.

Which Sensor Placements and Outcome Measures Are Used When Measuring Shock Impacts Using Wearable Sensor Technology in Sports?
The rapid growth in commercially available wearable sensors makes them easily accessible and affordable. However, for the results to be valid and reliable, the methods used for placement and data analysis are central. In the included studies, the dominant main outcome measure is derived from acceleration. There seems to be consensus that peak tibial acceleration (also referred to as tibial shock) can be used as a proxy for the impact forces experienced at the tibia and thus the vertical impact loading in running [85], as it has been shown to correlate strongly with vertical loading rates [86]. Comparably, for head impacts, peak linear acceleration is interpreted as detected head impacts. Many studies use a defined threshold for what would be interpreted as a head impact, most often set asSeattle, WA, USA g [41,42,51,54,58], but in three studies, it was even higher: ≥15 g [45,48] and ≥20 g [36]. Studies utilizing insoles base their outcomes on various measurements of force, vertical ground reaction force (vGRF), force impulse, and force/load distribution. Deceleration and braking are also reported as outcomes derived from both accelerometers and insoles.
In a recent review concerning measurements of tibial acceleration during running, Sheerin et al. [85] point out that different placements of accelerometers do not necessarily give comparable results. Furthermore, whether triaxial or uniaxial accelerometers are used will be of relevance, as accurate measurements from an uniaxial accelerometer is dependent on precise alignment along the long axis of the tibia [87]. Acceleration of the tibia occurs in three dimensions: axial, anteroposterior, and mediolateral. Triaxial systems will incorporate this and are thus less vulnerable to misinterpretations of the actual anterior acceleration of the tibia. In this review, six of the studies investigating foot strike in running with IMUs utilized triaxial accelerometers [34,35,37,52,57,60], two studies with the same first author used a combination of uni-and biaxial [43,44], whereas one study used uniaxial [47]. Furthermore, three studies utilized triaxial accelerometers, but with other primary outcomes than foot strike, and with accelerometers mounted on the trunk and/or several locations on the body [38,49,88]. With regard to the placement of those assessing foot strike as their primary outcome, there was considerable variation. Three studies had sensors placed at the anteromedial aspect of tibia [37,52,57], three studies had a combined placement on the heel counter of the shoe and anteromedial tibia [34,43,44], and the remaining three studies had sensors placed on the heel counter of shoe [35], on top of the shoe above the third metatarsal [47], and finally using a Lycra suit with multiple sensors on upper and lower extremities, trunk, and head [60]. This demonstrates a lack of consensus on sensor placement in the literature, which can make comparisons between studies very challenging. From the already mentioned review, Sheerin et al. also state that tibial acceleration measured by distally attached sensors gives higher values, which is a notion that is supported by Blazey et al. [87], who conclude that current evidence suggest IMU devices should be placed and fixated on the distal tibia. It should finally be noted that bone-mounted accelerometers have been shown to have the highest association with vertical load rates from force plates, with correlations of r = 0.97. In comparison, skin-mounted accelerometers, which are the relevant wearable for field situations and thus for the studies included in this material, have correlations of r = 0.70 [52].
With regard to head impacts, placement of the wearable sensors may be challenging, as they are prone to be broken by the impact. Wearable sensors for assessments of head impacts are available in several systems; they are embedded in instrumented helmets, headbands, mouthguards, and skin patches [89,90]. Of the ten studies included in the data material, skin patches (xPatch) were used in four studies [36,41,54,58], instrumented helmets (GForce Tracker and SpeedFlex/HIT) were used in five studies [36,40,42,45,51], a headband (SIM-G) was used in one study [48], and finally MVTrak, with a sensor placed in the ear canal, was used in one study [53] (number summarizes to eleven, as the study by Cortes et al. [36] utilized xPatch for females and GForce Tracker for males). Regardless of system, it is important to notice that the accelerations captured represent a combination of true head impacts-such as collisions or hitting the ground-and false detected impacts stemming from movement [39,41,91], e.g., change of direction, jumping, and decelerations. Even though impacts above 10 g, which is the most common threshold reported in the studies included in this review, are likely to be accrued by true impacts, several of the included studies show that this cannot be trusted unless confirmed by video. Linear and rotational acceleration magnitudes from lab studies have been shown to be over-predicted for sensor solutions in both skin patches and instrumented helmets [92]. False positive high acceleration impacts have further been revealed in field studies, and the importance of video confirmation of sensor-recorded events to remove false positives is in a recent systematic review by Patton et al. emphasized to avoid overestimation of head impact exposure [90]. Still, two-thirds of the included studies in their review did not include video. In this scoping review, numbers were higher, as seven out of ten included studies did include video confirmation, and several also emphasized the importance of doing so. Interestingly, Carey et al. [91] found the vast majority of high acceleration impacts, when defined as above 20 g, to be verified by video. This underlines the ambiguity of using 10 g as a threshold for head impacts, especially if not adding video to confirm events. The systematic review by Patton et al. [90] finally also states that even though the majority (74%) of the included studies used filtering algorithms, these remain inadequate. Thus, the studies included in this scoping review, as well as previous reviews, suggest that trusting wearable sensors as the only data source for assessments of head impacts in sports could imply an overestimation of potential harmful impacts, both in number and in severity.
Ground reaction force is an obvious variable of interest for mechanical analysis of eventual risk factors for injury, as well as for assessments of lower extremity load attenuation during sport activities that entail running, jumping, and landing. In this review, four studies utilizing pressure sensors were included. Of these, one study [59] used the Novel Pedar-X system to explore whether running speed affected plantar loads and to compare rearfoot versus mid-and forefoot strikers, without validating against a gold standard. The three other studies all validated against force plates [56,63] or an instrumented treadmill [55]. Stoggl and Martiner [56] validated the OpenGo sensor insole towards both a force plate and the PedarX assessing multiple motions, and found force impulses to be between 13 and 34% lower with OpenGo compared to force plates, emphasizing problems with force impact accuracy during short ground contact times. Seeley et al. [55] tested the accuracy of a nanocomposite piezoresponsive foam (NCPF) inserted into the running shoe under the insole, and comparisons with an instrumented treadmill revealed an error for predicted vGRF load rates between 22 and 29% for the NCPF. Finally, Seiberl et al. [63] analyzed the accuracy of Loadsol compared to a force plate during running and reported high precision of the sensors. However, the authors stated that insole devices are not accurate enough for highly dynamic GRF assessments, such as force rate.
Included in the data material are also three studies on running that have compared impact force assessments from accelerometry with force plate data. Ngho et al. [47] and Pogson et al. [49] both concluded that estimations of vGRF from accelerometer data using neural network modeling were promising, reporting small differences compared to force plate data. Sensors were placed on the trunk and on top of the shoe, respectively. Derie et al. [37] utilized machine learning to predict maximal vertical instantaneous loading rate (VILR) from triaxial accelerometers attached bilaterally to the tibia and concluded that multiple 3D tibial acceleration features gave a more accurate prediction of the VILR than the frequently used axial peak tibial acceleration, which is a single time discrete variable of tibial acceleration. Finally, two studies investigating landing impacts with validation from force plate data were identified, namely after somersault in gymnastics [31] and in ballet [62]. Whereas the first mentioned study concluded that external impact forces could not be estimated accurately based on accelerometer data, the second stated strong correlations between impact acceleration and peak vGRF.
The gold standard for GRF measurements are force platforms or treadmills instrumented with force sensors. However, such assessments are not available for the capture of complex movement patterns performed in the field, such as in team sports and downhill skiing. For running, an instrumented treadmill can provide the possibility to measure GRF from repeated foot strikes, simulating a long run. However, the environment will be controlled and thus not reflect variations posed by natural surfaces and terrain. Therefore, wearable technology could represent a considerable potential within this context. A systematic review from Ancillao et al. [93] found sensors that allow direct measurements of GRF-such as insoles, wearable load cells, or ad hoc designed pressure sensing devices-to be more reliable than GRF predicted from IMU data. This is confirmed in a recent review by Blazey et al. [87] who found instrumented insoles, in particular the Loadsol system, to offer a good in-field assessment tool. However, it is important to note that when sensors are worn under the foot, they compromise the foot-ground interaction, and the loads measured do not reflect the pressure absorbed by the tissue but rather the pressure on the device or the shoe to insole interface [87,93]. It is further emphasized that even if there is a correlation between predicted and directly measured GRF, it is difficult to estimate the absolute value of the peak force.

Which Knowledge Gaps Are Apparent in the Literature Regarding Wearable Sensor
Technology for Assessments of Shock Impacts within Sports?
As previously discussed, an observed knowledge gap in this scoping review is the lack of studies assessing the magnitude of single shock impacts during running, landing, abrupt changes of direction, and pivoting activities in different invasion sports and other team sports. As monitoring in these sports is performed under the purpose of managing total strain of training and competition to prevent overtraining and injuries, knowledge on the magnitude of each single shock impact would add significantly to what can be drawn from the quantification of multiple impacts and exposure over time. Likewise, measurements of single shock impacts would also be of interest in the track and field events such as the high jump, long jump, and triple jump, but no studies could be found. Furthermore, it was somewhat surprising that no studies could be found on winter sports with known single shock impacts of considerable force, such as in alpine skiing (especially the downhill and Super-G disciplines), freeskiing, snowboard, and ski jumping. As described before, studies on these sports utilizing wearable technology primarily have focused on technique and tactical dispositions. To fully understand the consequences of shock impacts during real-life training and competition in various and often demanding environments, laboratory studies alone are not sufficient.
The aim of this scoping review was to summarize what has been done previously and carry out data charting, and not to score study quality. Still, it must be mentioned that many of the included studies are characterized by a relatively low number of participants. Overall, 18 of the included studies had one to 20 participants, seven studies had 21-50 participants, five studies had between 51 and 100, and three studies had more than 100 participants. One study did not inform about the number of participants at all. In the studies with more than 50 participants, three studies assessed head impacts in invasion sports and two studies assessed running. Finally, in studies with more than 100 participants, two assessed head impacts in invasion sports and one study assessed running. Furthermore, as many as 20 out of the 34 included studies were based on a single intervention and/or methodological study design, with validation of the wearable sensors utilized as a primary goal. All the included studies assessing shock impacts in running fall under this study design category. This reveals that this field still should be characterized as being immature, and that measurement methods for shock impacts during real-life running using wearable sensors are not yet adequate as a stand-alone approach. Of the remaining designs, nine studies were prospective cohort studies on invasion and team sports where participants were followed over one or more seasons. Eight of these studies investigated head impacts [36,40,42,45,48,51,54,58] and one study investigated whole body impacts [39]. In addition, two studies investigated head impacts in a laboratory using video recordings to verify impacts measured by the wearable sensors [41,53]. From these studies, it seems that the number of head impact events may be overestimated from sensor data alone. This is explained from difficulties in separating real game-related head impacts, which are defined as direct contact to the head from collisions or blows, to impacts not affecting the head, rapid changes in direction, landings, or other abrupt movements. Head impacts may further be overestimated with regard to severity when trusting sensor data alone. Valid assessments of head impacts during game play in invasion and team sports are still dependent on confirmation by video to identify true-and potentially damaging-impacts. In sum, the included studies show that also in this context, wearable sensor technology at this point does not provide the necessary accuracy as a stand-alone method.
Finally, the numerous brands and types of wearable sensors utilized in different studies may pose a challenge when it comes to comparing protocols, field set-ups, and results. Additionally, there are several pitfalls related to the before mentioned lack of consensus regarding placement of sensors, as well as signal processing and filtering, that can lead to inaccurate interpretation of data [85]. Technological development in wearable sensor technology is fast, with a multitude of suppliers. Five years ago, Sperlich and Holmberg [94] pointed out that studies regarding the validity and reliability of wearable sensors had shown that many of the sensor technologies available on the market had questionable validity and reliability when used in various sport settings and populations. This review indicates that this concern is still valid for the particular scope of shock impacts, as so many of the studies are addressing validation. Additionally, no studies aiming to compare or validate different types or brands of accelerometers or IMUs up against each other were identified. Thus, caution should be made when considering the usability of wearable sensors of this type for different sports and movement contexts, as it is unknown whether there may be differences in accuracy and validity between different brands. IMUs are marketed as valuable tools for coaches and athletes. However, the findings from this scoping review support previous concerns that wearable sensors alone at present is not adequate to ensure valid and accurate data on shock impacts in the field. There is a need for future validation studies including larger populations, taking into account the specific characteristics of defined sport-specific tasks. Furthermore, validation studies should in addition to comparison with gold standard technologies include comparisons of different types and brands of sensors, sensor placement, as well as filtering and cut-off values for the detection and definition of thresholds to separate true impacts from acceleration caused by rapid movements.

Limitations
To our knowledge, this is the first scoping review to address the use of wearable sensor technology for assessments of shock impacts within the domains of sports and occupational settings. A generic limitation of the scoping review methodology is that it does not allow for a formal evaluation of study quality and the level of evidence, as the included papers represent a wide array of study designs and methods [95]. This is valid also for this study. The scoping review approach was chosen with intent to provide an overview of the body of literature within the two defined contexts and to identify apparent knowledge gaps, which may guide future initiatives for research within this field [96]. Sampling frequency or filtering of data were not included in the research questions of this study. In retrospect, it seems relevant to suggest that selected sampling frequencies and filtering techniques, including cut-off values, could be further investigated in future studies. The validity of wearable sensors in different contexts depends not only on placement of the sensors but also on the methods of data analysis. From the studies included in this review, this seems particularly relevant for complex movements with rapid changes of directions. The use of wearable sensors, especially within the sports domain, is expanding fast. Thus, we cannot eliminate the possibility that new studies accommodating the inclusion criteria may have been published between the search for eligible studies and publication of this review. Finally, we cannot eliminate the possibility that relevant papers might have been left out, due to the definitions and choice of terms in our search strategy.

Conclusions
The main aim of this scoping review was to provide an overview of existent methods for assessments of shock impacts based on wearable sensor technology within two domains: sports and occupational settings. No studies were found on occupational settings, and this is the most prominent knowledge gap identified in this study. For the sports domain, accelerometry was the dominant type of wearable sensor technology utilized for assessing shock impacts, interpreting peak acceleration as a proxy for impact. Of the 34 studies included, 28 assessed foot strike in running, head impacts in invasion and team sports, or different forms of jump landings or plyometric movements. The methodology of included studies revealed that there is a lack of consensus regarding sensor placement and interpretation of the results. Furthermore, the included studies that aimed at validation up against established gold standard methods support previous concerns that wearable sensors alone at the present time are not adequate to ensure valid and accurate data on shock impacts in the field. This advocates for high-quality research being needed to find the appropriate sensors and methodology to utilize the potential of measuring shock impacts in the field by wearable sensor technology.