Table 1.

Synthesis of results from included papers (n = 34).

Author(s) (Year)	Study Design	Study Aims	Main Outcome Measure of Relevance	Population (n, Mean Age ± SD)	Type(s) of Wearable Sensor Technology (Brand) and Placement of Sensor	Additional Technology for Validation	Key Findings of Relevance
RUNNING
Ching et al. (2018)	Cross-over study, test–retest	To compare the impact loading during distracted running before and after a two-week auditory feedback gait retraining program.	Peak positive acceleration (PPA)	Male and female (7/9) recreational runners (n = 16, 25.1 ± 7.9 years)	Tri-axial accelerometer (Maestro WB, 01DB-Stell, Limonest, France). IMUs placed on the heel counter of the shoe and at the anteromedial right tibia.	Instrumented treadmill	Real-time auditory feedback gait retraining is effective in impact loading reduction during distracted running. Runners after gait training do not benefit from augmented auditory feedback.
Colapietro et al. (2019)	Observational single intervention case-control study	To evaluate spatiotemporal, kinematic, and kinetic measures during the loading response of running using a wearable sensor during two 1600 m track runs at different intensities between recreational runners with and without chronic ankle instability.	Horizontal component of change in acceleration of the foot at initial contact (breaking g), vertical component of change in acceleration of the foot at initial contact (impact g)	Male and female (8/10) recreational runners (n = 18, 22.7 ± 4.7 years)	RunScribe™ Pro 2x running sensor with integrated triaxial accelerometer and gyroscope (RunScribe Labs, Half Moon Bay, CA). IMU placed on the heel counter of the shoe.		Altered running mechanics were demonstrated in individuals with CAI compared to healthy runners. The clinical utility of wearable sensors in this context should be noted.
Derie et al. (2020)	Methodology study	To evaluate the performance of accelerometer-based predictions of the maximal vertical instantaneous loading rate (VILR) with various machine learning models trained on data of 93 rearfoot runners.	Axial peak tibial acceleration (APTA), vertical instantaneous loading rate (VILR)	Male and female (55/38) recreational runners (n = 93, 35.3 ± 10.0 years)	Tri-axial accelerometers (LIS331, Sparfkun, Colorado, United States). Accelerometer attached bilaterally to the anteromedial side of tibia.	Force plates	Applying machine learning to multiple 3D tibial acceleration features results in a more accurate prediction of the VILR than the frequently used APTA, which is a single time-discrete variable of tibial acceleration. A subject-dependent model resulted in the most accurate predictions compared to subject-independent models.
Gageler et al. (2013)	Single-intervention test	To explore the feasibility of using simple kinematic parameters obtained from a single inertial sensor to determine the rate of breaking and stopping, and to observe which limbs are most affected by the various rates of deceleration. Explore the contributions of each segment in absorbing the forces associated with breaking.	Peak accelerations and shock attenuation	Experienced runners, recreational and professional (n = 3, gender and age not given)	Custom-built tri-axel ±8 g accelerometer. Sensor positioned around the middle to upper thoracic vertebrae. Custom built device containing a 100 g accelerometer. Sensor attached bilaterally to the distal fibulas.	Motion capture system with eight cameras (OptiTrack)	Peak ankle impact increases when the rate of deceleration increases. The rate of deceleration did not alter the peak torso acceleration. It is difficult to detect the rate of stopping using only peak impact parameters from an inertial sensor unit located on the upper torso.
Mitschke et al. (2017)	Single intervention test/validation of method	To investigate the influence of different inertial sensor sampling frequencies on kinematic, spatiotemporal and kinetic parameters during rearfoot running.	Peak tibial acceleration and peak heel acceleration	Male recreational runners (n = 21, 28.9 ± 10.8 years)	Uniaxial light-weight accelerometer (ADXL78, Analog Devices), placed at the medial aspect and mid location between malleolus and plateau of the right tibia, and an IMU, combining a biaxial accelerometer (ADXL278, Analog Devices, and a biaxial gyroscope (IDG-650, InvenSense), affixed to the heel cup of the right shoe.	Force plate	When investigating peak heel acceleration or parameters which are directly derived from the accelerometer signal (e.g., touchdown), sampling frequency should be as high as possible or at a minimum of 500 Hz. 200 Hz were required to calculate parameters accurately for peak tibial acceleration.
Mitschke et al. (2018)	Single intervention test	To investigate the influence of the accelerometer operating ranges (OR) on the accuracy of stride length, running velocity, and on peak tibial acceleration when reducing the OR stepwise from ± 70 g to ± 32, ± 16, and ± 8 g.	Peak tibial acceleration (operating ranges, where g is the acceleration of gravity)	Male recreational runners (n = 21, 24.4 ± 4.2 years)	Individually configured IMU combining a biaxial accelerometer (ADLX278, Analog devices with OR ± 70 g and a biaxial gyroscope (IDG-650, InvenSense), with OR ± 2000 deg/s. IMU affixed to the heel cup of the right shoe. Uniaxial accelerometer (ADLX78, Analog devices) with OR ± 70 g. Accelerometer attached at the medial aspect mid-distance between the malleolus and the plateau of the right tibia.		Operating ranges influenced the outcomes of all investigated parameters. The lower ORs were associated with an underestimation error for all biomechanical parameters, which increased noticeably with a decreasing OR. Accelerometers with a minimum OR of ± 32 g should be used to avoid inaccurate measurements.
Ngho et al. 2018)	Single intervention study/validation of method	To investigate the use of neural network model (NN) and accelerometer to estimate vertical ground reaction force (VGRF).	Foot forward acceleration, segmented and normalized from foot initial contact to end contact (threshold at 5 N) and used as input for NN	Male subjects (n = 7, 21.3 ± 0.5 years)	Inertial sensor measuring acceleration, angular velocity, and magnetic field (Opal, APDM Inc.) Only forward acceleration is used for analyses; thus, it is referred to as uniaxial accelerometer. Sensor placed on top of the running shoe above the third metatarsal.	Force plates	Using NN and a uniaxial accelerometer simplified the estimation of VGRF, reduced the computational requirement, and reduced the necessity of multiple wearable sensors to obtain relevant parameters.
Pogson et al. (2020)	Single intervention study/validation of method	To present an artificial neural network method to predict ground reaction force (GRF) time series from a single, commonly used trunk-mounted accelerometer.	Estimated GRF	Physically active male and female (10/5) team sport players (n = 15, 23 ± 1 years)	GPS-embedded accelerometer (MinimaxX S5, Catapult Innovations, Scoresby, Australia). Accelerometer worn in a tight-fitting vest on the back of the upper torso.	Force plate	GRF was predicted with an average r 2 of around 0.9 for the time series of each impact, and the method therefore offers a promising approach to estimate GRF in the field.
Reenalda et al. (2016)	Single intervention test	To present a measurement set-up based on IMUs, to perform a continuous 3D kinematic analysis of running technique during the course of an actual marathon to objectify changes in running mechanics.	Peak center of mass (COM) vertical displacement and acceleration	Well-trained male distance runners (n = 3, 38.7 ± 8.2 years)	Eight inertial magnetic measurement units (MTw, Xsens Technologies B.V., Enschede, the Netherlands) containing a 3D accelerometer, a 3D gyroscope, and a magnetometer. IMUs placed on trunk (sternum, just below the sternal angle), pelvis (on the sacral bone between left and right iliac spine), upper legs (on tibial tract, halfway iliac crest, and lateral condyle of the tibia), lower legs (at the lower third of the medial surface of the tibia), and feet.	GPS enabled watch	Peak COM acceleration (derived from the sacral sensor) increased in all three runners and might indicate higher loading rates, a reduction in shock absorption quality, and a higher impact on the body. The presented measurement technique allows for more in-depth study of the running mechanics outside the laboratory and of the effects of fatigue on running mechanics.
Ruder et al. (2019)	Single intervention study	To examine the relationship between foot strike patterns (and impacts across a marathon race: (1) compare landing impacts quantified by tibial shock, between rearfoot, midfoot, and forefoot strike (RFS, MFS, and FFS, respectively), (2) examine the relationship between TS and speed across FSP, (3) to investigate the effect of fatigue on impacts.	Tibial shock (peak tibial acceleration)	Male and female (119/103) females marathon participants (n = 222, 44.1 ± 10.8 years)	Accelerometer device (IMeasureU BlueThunder IMU, Auckland, New Zealand). Sensor placed on the anteromedial aspect of their right distal tibia.	Video camera (to detect FSP before running the marathon)	Findings suggest that MFS runners exhibit similar impacts as RFS, and both exhibit higher impacts than FFS. RFS and MFS both exhibit increasing impacts with increasing speed, whereas FFS runners do not. RFS and MFS runners are similar in their impact loading. An FFS pattern may be protective against increasing impacts with increasing speeds.
Seeley et al. (2020)	Single intervention test/validation of method	To test the accuracy of a nanocomposite piezoresponsive foam (NCPF) that can be inserted into the running shoe under the insole in predicting important characteristics of vertical ground reaction force (vGRF) during running at three different speeds.	Impact peak vGRF, active peak vGRF, max impact rate	Male and female (17/14) recreational runners (n = 31, 23 ± 3 years)	Right shoe instrumented with NCPF sensors and an accelerometer (Bosch Sensortec, Mount Prospect, IL, USA). NCPF sensors placed under the insole (toe, ball, arch, heel), accelerometer attached to shoelaces.	High-speed video, force instrumented treadmill	Percent error was relatively low for predicted vGRF impulse (2–7%), active peak vGRF (3–7%), and ground contact time (3–6%), but relatively high for predicted vGRF load rates (22–29%). For each response variable of interest, the most accurate models were subject-specific models.
Seiberl et al. (2018)	Single intervention test/validation of method	To compare the accuracy and precision of a new wireless insole force sensor for quantifying running-related kinetic parameters over an extended period of use to a gold standard device in a laboratory setting.	vGRF parameters (ground contact time, area under the force–time curve, active peak force, time to active peak force, and both positive and negative force rate)	Male and female (4/6) (sport students (n = 10, 21.8 ± 0.8 years)	Ergonomic and linearly sensitive capacitor-based sensor; Loadsol® insole (novel GmbH, Germany). Sensor placed inside shoes during running, under a tight-fitting sock for GRF comparative measurements.	Force plate	The mean bias of ground contact time, impulse, peak force, and time to peak ranged between 0.6% and 3.4%, demonstrating high accuracy of Loadsol® devices for these parameters. 95% of all measurement differences between insole and force plate measurements were less than 12%. Highly dynamic behavior of GRF, such as force rate, is not yet sufficiently resolved by the insole devices, which is likely explained by the low sampling rate.
Van den Berghe et al. (2019)	Test–retest	Reliability and validity data for axial and resultant peak tibial acceleration (PTA) along the speed range of over-ground endurance running is lacking. The study developed a wearable system to continuously measure 3D tibial acceleration and to detect PTAs in real time.	Tibial acceleration, ground reaction forces and detected PTAs	Male and female (7/6) uninjured rearfoot runners (n = 13, 33 ± 13 years)	MEMS tri-axial accelerometers (LIS331, Sparkfun, Niwot, CO, USA). Sensor placed on a lower leg alongside the distal anteromedial aspect, 8 cm above the medial malleolus.	Instrumented treadmill with force plate	The wearable system developed was able to continuously detect PTAs in real time and can be used for applications aiming at monitoring (e.g., before, during, and after an in-field intervention) the impact loading experienced in the time domain by a runner during real world locomotion.
Wei et al. (2020)	Single intervention test	To explore whether running speed affects plantar load, and to compare plantar loads between habitual rearfoot strike (RFS) and non-RFS (NRFS) runners under their preferred running speed.	Plantar loads (F)	Male distance runners (n = 66, 25.2 ± 3.5 years)	Novel Pedar-X system (Novel, Munich, Germany), each insole including 99 force sensors. Insole placed in right shoe.	Photoelectric timing system, high-speed video camera	55% of participants were verified as habitual RFS and 45% were verified as habitual NRFS. Habitual runners tend to adjust their contact area according to the running speed through midfoot and forefoot regions. RFS runners remain susceptible to high impact force on the heel and midfoot, and NRFS runners experience high impact force in the first metatarsal regions.
Wouda et al. (2018)	Single intervention study/validation of method	To examine the validity of a method (using artificial neural networks) to estimate sagittal knee joint angles and vertical ground reaction forces (vGRF) during running using an ambulatory minimal body-worn sensor setup.	Peak vGRF	Male experienced runners (n = 8, 25.1 ± 5.2 years)	Xsens MVN Link inertial motion capture system consisting of 17 IMUs (Xsens, Enschede, the Netherlands). Full body Lycra suit used for placement at both shoulders, upper arms, lower arms, hands, upper legs, lower legs, feet, head, sternum, and pelvis.	Motion capture system with high-speed cameras and instrumented treadmill with force plate	Sagittal knee kinematics and vGRF can be estimated using only three inertial sensors placed on the lower legs and pelvis. The peak vGRF are estimated with no significant differences with respect to the reference. Best performance can be obtained when the proposed approach is applied to a single subject.
INVASION AND TEAM CONTACT SPORTS (activity specified in first column)
Mihalik et al. (2016) Football	Prospective cohort study (throughout seasons over 8 years)	To investigate the clinical utility of head impact magnitude thresholds used by various commercially available head impact indicators to positively predict concussion among American football players.	Head impacts (impacts exceeding 10 g of peak linear acceleration)	Division I Football Bowl Subdivision college football players (n = 185, 19.2 ± 1.4 years)	Head Impact Telemetry (HIT) System; including linear acceleration, rotational acceleration, HIT severity profile (HITsp), head injury criterion (HIC), and Gadd severity index (Riddell Corp., Chicago, IL, USA). Sensors embedded in helmet.		The ability of a head impact indicator—used in isolation—to detect a concussive injury is minimal, even if it can accurately measure and report biomechanical outcomes. Injury thresholds used by existing head impact indicators cannot predict concussion when used in isolation.
Rose et al. (2018) Football	Longitudinal cohort study (throughout season)	To determine the association of repetitive subconcussive head impacts with functional outcomes in primary and high school tackle football players.	Head impacts ≥ 10 g	Primary school and high school football players (n = 112, 13.6 ± 2.9 years)	Riddell SpeedFlex or Speed helmet (Riddell, Rosemont, Illinois), containing the Riddell InSite Impact Response System, based on the Head Impact Telemetry (HIT) system Sensor placed between the player’s head and the helmet padding.		In youth tackle football, subconcussive head impacts sustained over the course of a single season may not be associated with neurocognitive functional outcomes. The absence of a significant association may reflect the relatively short follow-up interval and signals the need for studies across multiple seasons.
McIntosh et al. (2018) Football	Laboratory tests and prospective observational study (over two games)	To assess the utility, functionality, and wearability of the X-Patch® as a measurement tool to study head impact exposure in sports without helmets, using Australian Football as an exemplar sport.	Head impacts (peak linear acceleration ≥ 10 g).	Male and female (24/29) amateur level adult Australian football players (recruited n = 97, included in analysis n = 53, 26.0 ± 2.0 years)	X-Patch (X2 Biosystems, Seattle, WA, USA). Sensor placed over the mastoid process.	Video analysis	The X-Patch had limitations in two distinct areas: accuracy (laboratory tests) and validity (field tests). The potential errors were considerable and could result in sizable misreporting of the head impact incidence rates. Use of the current X-Patch® devices should be limited to research only and in conjunction with video analysis.
Muise et al. (2016) Football	Observational study (throughout season)	To examine how frequently, and to what magnitude, Canadian University football players get hit in training camp and how this compares to practices and games in regular season.	Head accelerations greater than 15 g	Male players in Canadian Interuniversity Sport (CIS) football (n = 47, age not given)	GForce Trackers (GFT) (Artaflex, Markham, Ontario, Canada). Sensor placed inside football helmet, to the right side of the crown.		Data from 20,950 impacts revealed that games were associated with significantly larger magnitudes and frequencies than either training camp or practices, but that training camp was associated with significantly greater magnitudes and frequencies than in-season practices. In addition, positional differences existed.
Bruce et al. (2019) Basketball	Single intervention study	To quantify the influence of basketball court surface construction and shoe midsole stiffness on ground reaction force (GRF), lower-extremity joint work, impact, and impact attenuation during countermovement jump landings.	Resultant peak acceleration tibia and head impact attenuation	Male collegiate and high school basketball players (n = 29, 19.1 ± 3.3 years)	Uniaxial and tri-axial piezoelectric accelerometers (ADXL78 and ADXL1002, Analog Devices, Inc., Norwood, MA, USA). ADXL78 mounted to the forehead and ADXL1002 mounted to the tibial tuberosity.	Force plates, high-speed cameras and a Vertec Jump trainerTM	Shoe stiffness and surface had minimal effects on parameters associated with impact during countermovement jump landings. Landing in a more compliant shoe reduced peak ankle moment and tibial impact acceleration. Results for tibial impact acceleration were inconsistent.
Cortes et al. (2017) Lacrosse	Prospective cohort study (throughout season)	To utilize video analysis to verify head impact events recorded by wearable sensors and describe the respective frequency and magnitude.	Head impacts with linear acceleration ≥ 20 g; leading to recording of impact	Male and female (30/35) high school lacrosse players (n = 65, 16.4 ± 1.3 years)	Females: X-patch sensors including a triaxial accelerometer and a gyroscope (X2 Biosystems, Seattle, WA, USA). Sensor placed at the right mastoid process. Males: GForce Tracker including a tri-axial accelerometer (GForceTracker Inc.). Sensor affixed to the inside crown of the helmet.	Video recordings	65% and 32% of all head impacts recorded during boys’ and girls’ lacrosse game play were verified as true game play-related head impacts by video analysis, respectively. Results suggest that existing wearable sensor technologies may substantially overestimate head impact events.
Kelshaw et al. (2018) Lacrosse	Prospective cohort study (throughout season)	To assess the effects of isometric cervical muscle strength (ICMS) on head impact kinematics (HIK) in high school boys’ lacrosse, and to investigate the relationship between cervical anthropometric measures (CAM) and ICMS.	Head impacts (linear acceleration (g) and rotational velocity in degrees per second)	Male high school varsity lacrosse players (n = 15, 16.5 ± 1.33 years)	IMU with tri-axial accelerometer and gyroscope (GForceTracker™, Markham, ON, Canada). Sensor adhered to the inside crown of the helmet.	Handheld dynamometer to measure ICMS, high-definition camera	Thirteen of the participants sustained game-related impacts that were confirmed on video. A total of 367 impacts were confirmed using video analysis for the 13 participants. ICMS did not affect HIK, and CAM did not approximate ICMS. Findings suggest that greater ICMS alone may not mitigate HIK in collision sports.
Vollavahn et al. (2018) Lacrosse	Prospective cohort study (throughout season)	To establish the frequency of head impacts across impact mechanism, and to determine differences in linear and rotational head impact accelerations according to impact mechanism in NCAA Division III men’s lacrosse athletes.	Head impacts, threshold 10 g of linear acceleration	Male athletes in NCAA Division III men’s lacrosse (n = 11, 20.9 ± 1.3 years)	xPatch sensors (X2 Biosystems, Seattle, WA, USA) containing a triaxial accelerometer and rotational gyroscope. Sensor placed over the participants’ right mastoid process.	Video camera	A total of 167 head impacts were successfully verified and coded with a mechanism using video footage during 542 total participant exposures. The highest incidence rate was head to body, and the lowest was head to ball. The study failed to find differences in head impact magnitude depending on the mechanism that caused the impact.
Grainger et al. (2018) Rugby	Longitudinal observational cohort study (throughout season)	To compare the absolute and relative number of impacts between nine positional groups in rugby union.	Impacts defined from values above 2 g in a 0.1 s period (of which 10 g impact classifications are likely to be accrued by collisions)	Professional male rugby players (n = 38, 26.4 ± 4.7 years)	GPS units with integrated triaxial accelerometers (StatSports Viper, Northern Ireland). Incorporated in jerseys, on the thoracic spine between the scapulae.		The frequency and magnitude of impacts experienced by positional groups vary. Inertial sensor impacts encountered during match play are likely a combination of “real physical impacts” from collisions and those accrued from movement tasks (deceleration, landings, and changes of direction). It is important to assess the total inertial sensor impact values accrued during match play with caution.
Patton et al. (2020) Soccer	Prospective cohort study (over two seasons)	To (1) identify the percentage of video-confirmed events recorded by headband-mounted sensors through video analysis; (2) compare video-confirmed events with the classification by the manufacturer filtering algorithms; and (3) quantify and compare the kinematics of true- and false-positive events.	Resultant linear acceleration and angular velocity	Male and female (49/23) adolescent varsity soccer players (n = 72, age not given)	Triaxial gyroscope for measurement of angular velocity and a high- and low-g triaxial accelerometer for measurement of linear acceleration (SIM-G), trigger threshold 16 g. Sensor mounted in a neoprene headband and positioned just above the greater occipital protuberance.	High-definition video camera	Of the 1893 sensor-recorded events in the final dataset, video confirmation revealed that 1316 (70%) were impact events, 396 (21%) were trivial events, and 181 (10%) were non-events. Percentages of video-confirmed impact events, trivial events, and non-events varied by sex. Current manufacturer filtering algorithms and magnitude thresholds are ineffective at correctly classifying sensor-recorded events.
Sandmo et al. (2019) Soccer	Descriptive laboratory study/single intervention study	To test the validity of an in-ear sensor for quantifying head impacts in youth soccer.	Head impacts (peak linear acceleration (PLA), peak rotational acceleration (PRA), and peak rotational velocity (PRV))	Male youth soccer players at the regional elite level (n = 6, 15.3 ± 0.3 years)	Sensor device MV1 (MVTrak). Placement of sensor in the left external ear canal.	Digital video cameras	The in-ear sensor displayed considerable random error and substantially overestimated head impact exposure. Despite the sensor’s excellent on-field accuracy for discriminating headings from other accelerative events, there is a need for secondary means of verification (e.g., video analysis) in real-life settings.
Saunders et al. (2020) Soccer	Prospective cohort study (throughout season)	To compare head impact magnitude and frequency between men’s and women’s intercollegiate soccer players based on head impact mechanism.	Head impacts ≥ 10 g (linear acceleration (g) and rotational acceleration (deg/s2)	Male and female (12/16) intercollegiate soccer players (n = 28, 20.1 ± 1.1 years)	xPatch (X2 Biosystems, Seattle, WA, USA) head impact sensors. Contains a triaxial accelerometer and gyroscope. Sensor placed over the participants’ right mastoid process.	Video camera	Only head impacts that could be clearly seen on video were included. Most of the head impacts recorded in the current study were below 20 g. Men’s soccer athletes sustain head impacts more frequently than women. Women had the highest head impact frequency when heading a soccer ball, while men were most likely to sustain head to body contact.
PLYOMETRIC ACTIVITIES AND LANDINGS (activity specified in first column)
Nagano et al. (2018) Badminton	Observational study (two games)	To elucidate the movements requiring greater trunk accelerations and its frequencies during badminton games and compare the acceleration components among such movements.	Resultant trunk acceleration >4 g	Female badminton players (n = 10, 15.8 ± 1.0 years)	Triaxial accelerometer (SS-WS1201, Sports Sensing, Fukuoka, Japan). Accelerometer secured to the upper back using a vest.	Two digital video cameras	The movements generating greater trunk acceleration were running during an underhand stroke on the dominant hand side leg, landing after an overhand stroke on the dominant and non-dominant hand side legs, and cutting from a split step using the dominant and non-dominant hand side legs.
Almonroeder et al. (2019) Ballet	Single intervention study/validation of method	To compare impact forces in ballet measured by a wearable accelerometer to loading rates of GRF during a common ballet maneuver involving jumping.	Peak impact acceleration and peak vertical GRF	Experienced female ballet dancers (n = 15, 18 ± 4.5 years)	Tri-axial accelerometer/IMU (Shimmer3 IMU). Sensor placed on the left anterior superior iliac spine, with the vertical axis of the device aligned with the long axis of the trunk when standing.	Force plates	Strong positive correlations were found between impact accelerations and peak vertical GRF, indicating that a wearable accelerometer can provide a means of indirectly capturing GRF features in ballet dancers during landing.
Bradshaw et al. (2020) Gymnastics	Single intervention study/validation of method	To examine measurement agreement between resultant peak force with vertical peak force and peak resultant deceleration of backward somersault landings.	Peak resultant and vertical ground reaction force (GRF), peak resultant deceleration	Competing female artistic gymnasts (n = 7, 10–15 (± not given) years)	Iso-inertial measurement unit/ IMU (iMeasureU, Auckland, New Zealand) IMU placed on upper back (T2).	Force plates	Moderate measurement agreement was revealed between the peak resultant force and peak deceleration measures, indicating that the external impact forces for backward somersault landings cannot be adequately estimated using an IMU/accelerometer placed on the upper back. Alternative placement of IMUs needs to be explored (ex. distal tibia, L5).
Campbell et al. (2020) Gymnastics	Single intervention study/validation of method	To determine the most appropriate filter cut-off for acceleration- and force–time data when measuring peak resultant acceleration (PRA) and ground reaction force (PRGRF) during landings after standing backward handspring and backward somersault from a height.	Peak resultant acceleration	Male and female (8/8) competitive artistic gymnasts (n = 16, 14.1 ± 3.6 years)	Triaxial IMU (iMeasureU, Auckland, New Zealand) IMUs placed on upper back (T2), lower back, both sides of the posterior superior iliac spine and forearms for backward handspring; on upper and lower back and bilaterally in tibia for backward somersault.	Force plate	For applied sports settings, no filtering is needed. However, a minimum cut-off of 85 Hz should be implemented for research purposes investigating pooled data from a large number of participants.
Stöggl et al. (2017) Multiple motions	Single intervention test/validation of method	To validate the accuracy of the OpenGo sensor insole, compared to the AMTI force plate system and the gold-standard sensor insole system, PedarX, during walking, running, jumping, body balance, and imitation motions.	Impulse of force, maximal force and mean force (g-range ± 8 g)	Male and female (14/2) sport science students (n = 16, 31 ± 10 years)	Sensor insoles OpenGo (Moticon GmbH, Munich, Germany). Consists of two sensor insoles (containing 13 capacitive sensors each) that measure the plantar pressure distribution and the acceleration in three dimensions in space. PedarX Mobile System (Novel GmbH, Munich, Germany) measures pressure distribution and used for comparison. Both insoles sandwiched between the foot and the inside of the shoes.	Force plate	The OpenGo system demonstrated almost perfect agreement with force plate data for detection of cycle characteristics and temporal parameters during gait and jumping tasks. Force impulses were 13–34% lower with OpenGo when compared to AMTI. During fast motions, with high force and impact, OpenGo provided lower force and latency in force kinetics. Thus, very short ground contact times with force impacts cannot be determined accurately by the OpenGo system.
Ziebart et al. (2017) Multiple motions	Single intervention study/validation of method	To examine how system characteristics, such as operating range and sampling rate, influence the measurement of peak impact loads by commercial activity monitoring systems as compared to a laboratory-grade criterion standard accelerometer.	Peak impact loads (acceleration gmax)	Male and female (5/7) (n = 12, 24.1 ± 2.6 years)	Three tri-axial accelerometers; a criterion standard laboratory-grade unit (Endevco 7267A) and two systems primarily used for activity monitoring (ActiGraph GT3X+, GCDC X6-2mini). All three accelerometers were affixed to one another using double-sided tape, secured to the participant’s left anterior superior iliac crest (ASIS).		Accelerometers designed for activity monitoring underestimated peak impact magnitude by up to 35%. Underestimation error was greater for tasks with greater impact magnitudes. Both the type and intensity of activity should be considered when selecting an accelerometer for characterizing impact events. In addition, caution may be warranted when comparing impact magnitudes from studies that use different accelerometers.
Ross et al. (2016) Conference abstract Snowboard	Not given	To report on the progress of a four-stage research program to develop an athlete tracking system suitable for use by snowboard athletes, where stage 2 investigates the relationship between body mounted accelerometers and landing impacts.	Not given	Not given	IMUs containing an accelerometer, gyroscope, and magnetometer sensor (The OptimEye (Catapult, Australia) and IMeasureU (I Measure U, New Zealand)). Sensor placement not given.	Force plate	Both IMUs showed similar correspondence with the gold standard (force plate). Findings will be used to refine data collection and processing techniques for stage four of the program; implementation and validation of an on-snow athlete tracking system.