More than 90% of regular runners use a Global Positioning System (GPS)-equipped watch or some other run tracking device (1). By 2022, 105 million wearable fitness devices are expected to be sold, totaling in excess of US $3.33 billion in revenue (2). Runners, coaches, and clinicians use wearable devices primarily to track training loads and assess running biomechanics in the hope to: (a) improve running performance (3) and/or (b) reduce risk of experiencing or assist in treating running-related injuries (3,4).
Despite the widespread use, fewer than 10% of commercially available wearable devices are validated against an accepted gold standard, suggesting that current enthusiasm for these devices should be tempered (3). For instance, ViPerform™ (DorsaVi™, Melbourne, Australia) estimates the peak vertical ground reaction force (GRF) during running via a shin-mounted accelerometer. Compared with the gold standard force plate, the ViPerform™ estimates peak vertical GRF during running with an absolute error of approximately 400N (5). Considering that 400N represents more than 50% body weight of a typical runner (assumed mass of 60 kg to 70 kg) and is approximately 20% of the expected peak vertical GRF of running (6), the ViPerform™ lacks the ability to precisely track training loads in runners.
Still, wearables provide the capability of moving data collections outdoors, that is, in-field, and may provide important insight into the relationships between validated metrics and markers of performance, fatigue, and injury risks. Ruder and colleagues (7) assessed tibial shock via a wireless inertial measurement unit (IMU) attached to the distal anteromedial tibia in 222 runners during the 2016 Boston Marathon. After controlling for changes in running speed, tibial shock did not change across the course of the marathon, regardless of footstrike pattern (7). Intriguingly, these findings conflict with controlled laboratory studies that found tibial shock increased from the beginning to the end of exhaustive treadmill runs (8,9). The data of Ruder et al. (7) suggest that wearables can provide important insights into in-field running biomechanics that were not possible previously in laboratory-based investigations.
To maintain pace with the rapidly evolving field of the use of wearables in runners, this review provides an update on the state of the literature with a particular focus on literature published in the past two years. Case studies illustrate the use of wearable data in development or monitoring of running programs. For the purposes of this review, “wearable device” was operationally defined as a device that can be attached to the runner, shoe, or garment — or is a smartphone application.
Tracking Training Loads in the Endurance Runner: The Role of Wearables
External training load is an objective measure of work or applied biomechanical loads experienced by the runner (10). In contrast, the runner's psychobiological response to a workout, for example, session rate of perceived exertion and heart rate, is defined as the internal training load (10) and is beyond the focus of this review.
The quantification of external training loads is the focus of the majority of recent developments in wearable devices. Examples of devices that assess external training loads include GPS-equipped running watches to measure speed and distance, accelerometers, and IMU to quantify spatiotemporal metrics, segmental accelerations, and estimate the vertical GRF, and instrumented insoles to assess plantar pressure distribution and quantify the normal force (3,4,11). These types of devices are popular, relatively inexpensive, and easy to use. Typically, commercial wearable devices collect data, upload data to a cloud server that is then processed by proprietary algorithms, and generate user reports via a web-based platform.
Historically, training diaries have been used by runners to track external training loads (12). While training diaries are inexpensive and easy to administer, they are prone to self-report and recall errors (12). When compared with GPS running watches, self-reported training volume may have up to a 36% error (13), strongly suggesting that training diaries are not a valid means to track training loads. A 2- to 4-wk lag is present between a training load error and the subsequent injury (14), underscoring the importance of objectively tracking and assessing training loads with as little delay as possible upon completion of a training session. Theoretically, wearable devices can objectively track training loads in real-time or near real-time and, through the supporting algorithms, identify training load errors prior to the occurrence of running-related injury.
The ability to quantify forces or stress at the tissue level would prove invaluable in the prescription of training loads, particularly in the rehabilitation of runners. For instance, quantifying Achilles tendon forces in a runner recovering from Achilles tendinopathy may provide critical guidance during the return to run phase of rehabilitation. Implantation of strain gauges within the Achilles tendon can quantify in vivo tissue loads (15), yet these devices are invasive and likely alter running mechanics. Alternatively, Achilles tendon forces during running can be readily estimated, but only in laboratories via musculoskeletal modeling techniques which requires modern motion capture technologies and a high level of expertise. Clearly, the in vivo quantification of tissue level loads is only feasible in research settings (16), requiring the development of surrogate metrics for the quantification of training loads in the field through the use of wearable devices. While high-quality wearable devices can quantify a number of loading parameters with a high degree of accuracy, for example, step counts, it should be evident that loads measured by wearables, such as insole pressures, are the loads applied to the device and are not tissue-level loads. To date, wearable technology is not a valid tool to assess tissue-level loads, but several studies have validated wearable technology to quantify global loads experienced by the runner, such as vertical GRFs (11,17).
Loads experienced during dynamic tasks, such as walking, running, and jumping, can be measured using wrist-worn accelerometry (e.g., GENEActiv™), devices typically used to quantify physical activity. Specifically, evidence shows high correlations between the peak accelerations obtained from wrist-worn accelerometers and peak vertical GRF, resultant GRF, and peak loading rates (18). Not only can such devices be used to unobtrusively quantify the load experienced by a runner, but they also can be used to identify running training days and estimate traditional external load metrics (e.g., miles, duration) (19). These advancements would reduce the reliance on self-reported training logs. Additionally, the work by Stiles and colleagues (19) used high-resolution (100 Hz) accelerometer data, which is ten times higher than most commercial GPS devices used to quantify external workload (20) an important factor to consider regarding the accuracy of estimated external load metrics. Finally, by design, a GENEActiv™ focuses on all physical activity undertaken by individuals rather than during one-off training sessions. This may actually provide a “true” external load picture, as it will allow all activity undertaken by an individual to be analyzed. Initial work by Gruber and colleagues (21) has started to quantify the impact of sedentary time on injury risk, but further work is required to consider this within the context of monitoring training.
The Use of Wearable Devices in Optimizing Runner Performance
Traditional external workload metrics, such as running distance (e.g., miles), duration (e.g., hours:minutes), and speed (e.g., minutes per mile), can be tracked during training blocks and in the lead up to a competition by running watches equipped with GPS technologies. Importantly, these running watches are valid assessments of distance and speed during linear and moderate curvilinear running typical of endurance running, provided the GPS sampling frequency is 1 Hz or higher (4,22). Based on a previous meta-analysis (23), appropriate precompetition tapering whereby training volume is exponentially reduced can maximize performance gains. In fact, precompetition tapering can be predicted based on a runner's regular running training content (24). For example, the authors found that higher weekly continuous running volume during regular training was related to larger reductions in continuous running volume during the taper phase (r = −0.70). This exciting avenue of potentially providing prescribed and objective precompetition tapering content could be implemented using wearables, but to date has not been developed.
Traditional external workload metrics (e.g., miles, duration, and velocity) only provide part of the workload picture, providing no information regarding how a runner has covered a specific distance. Runners in the same training group may run the same distances at the same pace, but will likely experience vastly different internal biomechanical loads, reflecting the complex interaction between running biomechanics, structure, and tissue qualities (16). Further, running over different terrains, such as hills or different surfaces, may result in large differences in applied biomechanical loads compared with level running on pavement (Fig. 1) (25). Several wearable devices, including accelerometers, inertial sensors, and insole pressure measurement systems, provide running technique information (e.g., cadence, stride length, contact time). These devices have the potential to generate a greater understanding of the external loads experienced by the lower limb during running. The combination of traditional metrics and specific running mechanics may be beneficial for both performance and injury risk. As an example, we will focus on discussing running velocity, cadence, and stride length.
Commercially available, accelerometer-equipped GPS running watches accurately measure running cadence and running velocity concurrently (20,26,27). Cadence (i.e., step or stride frequency) and step/stride length determine running velocity. Runners self-optimize cadence and stride length to minimize their metabolic cost of running (28–31), allowing them to run for longer at the same velocity or to run at higher velocities over a given distance. In particular, trained runners adopt a cadence and stride length that is near to their mathematically derived optimal values for these characteristics (31–33), while untrained runners are further away from their mathematically derived optimal values (8% vs 3% for untrained and trained runners, respectively) (32). Recently, it has been suggested that trained runners can operate within a window of different stride lengths (self-selected to self-selected minus 3%) without compromising their metabolic cost (28). This optimal window allows for potential natural variation in running gait, which runners can easily track with wearables. While the majority of studies within the field of economical running focus on measuring oxygen consumption (29–31,33), which requires expensive equipment and technical expertise, de Ruiter and colleagues (32) validated the use of heart rate as a surrogate measure for economical running. Given the plethora of heart rate monitors on the market, determining a runner's optimal cadence and stride length is accessible for coaches and practitioners.
Recent advances in inertial measurement unit technologies have growing promise in quantifying running kinematics. Still, recently published studies suggest that accurately estimating running kinematics with wearable devices is not trivial, largely due to sensor noise and signal drift (34). For instance, a shoe-mounted inertial measurement unit resulted in large systematic and random disagreement with a gold standard motion capture system for such basic running kinematics as foot strike angle and eversion excursion (35). Wouda and colleagues (36) were able to obtain improved criterion validity for sagittal plane knee kinematics (root mean square errors <5°) by using an array of 17 inertial measurement units worn by runners, in combination with a machine learning approach. Dorschky and colleagues (37) further refined estimates of sagittal plane hip, knee, and ankle running kinematics by combining segment positions and orientations obtained from an array of inertial measurement units worn by subjects with global optimization and musculoskeletal modeling. Errors were 5 degrees or less for sagittal plane estimates of ankle and knee kinematics, with a slightly higher error of approximately 9 degrees for sagittal plane hip kinematics (34). Intriguingly, Dorschky et al. (34) also estimated sagittal plane hip, knee, and ankle moments with excellent correlations (r ≥ 0.90) compared with gold standard 3-D motion capture, suggesting that estimates of joint specific loads may be possible during running in-field. Overall, these findings suggest that most estimates of running kinematics via wearable devices can be done relatively accurately but require advanced computational methods that are not yet currently commercially available.
As such, very few kinematics that can be readily determined using wearable sensors have strong relationships with running economy or injury. For instance, advancements have made it possible to assess the foot strike pattern used by a runner through an instrumented insole (38). Yet, a recent review recommended that altering foot strike is unlikely to reduce metabolic cost (28), and in fact, rearfoot strikers who adopt a forefoot strike could experience an increase in metabolic cost (39). Furthermore, adopting a forefoot strike does not appear to reduce overall injury risk in runners (40). The difficulty for wearable sensors to provide performance- and/or economy-relevant kinematic information lies in the challenging nature of being able to accurately and reliably determine toe-off, that is, the instant that the foot leaves the ground. Kinematics associated with this phase of the gait cycle have shown strong associations with running economy. In particular, generating less knee extension and less ankle plantarflexion is advocated as an economical running style (28). Such a style would enable a runner to produce greater propulsive force (30), through the leg extensor muscles operating at a more favorable position on the force-length curve. Algorithm developments are warranted to improve the utility of wearable technology for kinematics linked to performance and/or economy.
Case-Study Performance Perspective in the Use of Wearables in Runners
Determining optimal cadence and stride length can be easily undertaken with a runner and could be determined over different training phases and/or over a range of velocities to ascertain whether the runner has a stride length or cadence dominant strategy. For example, as running velocity increases does the runner self-select increases in cadence to produce the increase in velocity rather than an increase in stride length. Sprinters display a dominant strategy, which varies between individuals (41) and for endurance runners able to reach relatively fast velocities this also may be the case. By utilizing both heart rate and cadence/stride length data from wearable devices both the runner's dominant strategy and level of optimization could be identified. The same principle can also be applied to ground contact time, which has recently been shown to be self-optimized within trained runners (42).
A treadmill provides the easiest way to determine a runner's optimal cadence, as running velocity can be kept constant. A runner should run for a minimum of 4 to 6 min with their preferred cadence, noting down their heart rate and cadence during the final minute. Runners need to reach a steady-state, which can be identified by their heart rate stabilizing. Then, they should systematically increase and decrease their cadence, typically by ±5% and ±10%, each time running for 4 to 6 min and noting down their heart rate and cadence during the final minute. For example, a runner with a preferred cadence of 160 steps per minute should aim to run at 144, 152, 168, and 176 steps per minute. The data can then be entered into a bespoke spreadsheet (available upon request from author IM) or free software (42) to determine their mathematically optimal cadence (Fig. 2). For trained runners, the identification of an optimal cadence could potentially reduce metabolic cost by around 3 mLO2·kg−1·km−1 (31–33). For untrained runners, the effect is likely to be more pronounced as they are further away from optimal than trained runners (32). Therefore, this simple assessment could expedite an untrained runner's natural physiological and mechanical adaptations, while for trained runners, it could provide an avenue to hone their cadence with minor adjustments.
Injury Perspective of External Training Load
Quantifying applied biomechanical loads through the use of validated wearables may prove to be highly insightful in running epidemiology studies and provide guidance in the return to run process for the injured runner. Indeed, the highly repetitive biomechanical loads of running result in microdamage to local tissues (43) and, provided there is adequate rest and energy availability, tissues will remodel and adapt (43,44). However, biomechanical loads that exceed the rate or ability to adapt may result in a relative overuse injury (43,44). Please see Edwards (44) for an excellent review on the role of biomechanical loads in relative overuse injuries.
A large increase in the number of running epidemiology studies that use wearable devices is expected as devices improve and become more economically feasible. One such study (45) used a combination of a GPS running watch and a smartphone equipped with a mobile application (Help2Run™, Denmark) to determine if excessive increases in running volume or running intensity had differential risks of sustaining a running-related injury in 447 healthy, recreational runners. If a runner ran too fast or too slow during training sessions, the wearable device provided haptic feedback to cue the runner to adjust their running speeds accordingly. Interestingly, there were no significant differences in injury incidence between runners who were randomized to a training program that emphasized progressions in intensity versus those who were randomized to a training program based on volume progressions (45). This study was the first to provide real-time feedback on running pace, while tracking running volume and pace via GPS-technologies, all while assessing a large number of runners using a cloud-based system (Help2Run™). The large sample size in this study is noteworthy and would not be possible without the use of the wearable devices. The ability of wearable devices to capture massive sample sizes in running epidemiology studies will bolster statistical power in future studies to provide insight into less common and poorly understood running-related injuries, such as femoral stress fractures (4). One such ongoing study aims to collect data from 20,000 runners by combining self-reported injury data and running metrics captured by each runner's accelerometer-equipped GPS running watch (46).
Quantifying exposure to loading cycles, that is, steps, in runners may provide a means to estimate applied biomechanical loads. Step counts and/or run cadence is readily calculated by most running watches. Cadence is a fairly stable measure across endurance-paced running speeds (47) and may be an important metric when assessing risk of running-related injury. In a theoretical model, a longer step length while maintaining running speed (reduced run cadence) increased the risk of a tibial bone stress injury (48). A lower cadence at a fixed speed also increases total mechanical energy absorbed and generated per step (49), suggesting that using step lengths to monitor total biomechanical loading on a runner may have promise. To investigate the proposed relationship between run cadence and injury risk, Luedke and colleagues (50) measured cadence via a wearable in high school cross country runners. High school cross country runners who fell in the lowest quartile of cadence, that is, longer step length, were more than five times more likely (odds ratio, 5.85; 95% confidence interval, 1.1–32.1) to experience shin pain than runners in the top quartile for cadence (50). Running with a low cadence at a given run speed is associated with higher knee loads and higher vertical GRF (51) and higher braking GRF (52), which are biomechanics previously associated with running-related injuries.
Feedback to cue an increase in running cadence reduces certain lower extremity biomechanics associated with running-related injury (48,49,53–57). For instance, running with a 5% increase in running cadence compared to the preferred cadence reduces pressure on the plantar surface of the heel by over 565 body weights × s (BW × s) per mile run despite the increased number of steps that would be required by the increased step frequency (57). Running with a 5% to 10% increase in run cadence also reduces peak patellofemoral joint contact forces by 11% to 22% (53,58,59). Modern GPS running watches equipped with an accelerometer can provide visual, auditory, or haptic feedback when a runner falls outside prescribed targets for running cadence, enabling in-field feedback on running cadence (26,55). Please see case study following this section for recommended best practices in incorporating wearable devices in a gait retraining program for an injured runner.
Wearables that provide surrogate measures of GRF may provide important insights in external training loads in runners (4,43). Kiernan and colleagues (17) estimated the peak vertical GRF during running via a triaxial accelerometer mounted on the right iliac crest in nine NCAA Division I track runners over 419 training sessions. Compared with a gold standard force plate, the accelerometer was able to estimate peak vertical GRF within 8.3% ± 3.7% (17). The peak vertical GRF from each footstrike were weighted to the ninth power to model the stressed-life plot of soft tissue, that is, S-N curve, and subsequently summed to provide a mean weighted cumulative load per training session that accounted for steps per training session. Compared with uninjured runners, runners who developed a running-related injury had significantly greater peak vertical GRF and greater weighted cumulative load per run (17). These findings are intriguing since the runners were all members of the same track team and presumably had similar workloads in terms of run distance and pace, yet experienced differing vertical GRF. These methods suggest that using wearables to track the estimated runner-specific peak vertical GRF may provide clinically relevant insights into biomechanical loads applied globally to a runner's body.
Peak-positive tibial acceleration (“tibial shock”) in runners, measured via an accelerometer mounted on the distal tibia, is suggested to provide insight into tibial stress injury risk (6). Tibial shock during running is positively correlated with tibial bone loads during the loading response of running in a cadaveric model (60). Tibial shock is also correlated with average loading rate of the vertical GRF (61), also thought to be a risk factor for tibial stress injury in runners (6). Feedback on tibial shock, that is, gait retraining, results in reductions in both tibial shock and vertical loading rates (62), although the effects on tibial bone forces are unknown. Commercially available IMU are of sufficiently small mass (3 g), appropriate resolution (≥ ± 20 g), with sampling frequencies ≥500 Hz and with the capability of data logging of up to 12 h to enable the assessment of tibial shock during in-field runs (4). IMeasureU™ (IMeasureU.com; Vicon, Oxford, UK), for instance, is a tibial mounted IMU that quantifies tibial shock logged over the course of a run. Tibial acceleration data from the IMU are then entered into a daily load stimulus algorithm (63) which accounts for the rate of skeletal remodeling in response to biomechanical loading to provide a metric of “bone load” per training session (64). By factoring chronic bone loads into an acute/chronic workload ratio calculation (64), guidance can be provided for the prescription of training loads. Importantly, the use of metrics of estimated bone load has not been validated in observational or interventional studies. For instance, it is unknown if using daily load stimulus data to prescribe training loads actually reduces the risk of experiencing a tibial bone stress injury in runners.
Equating GRF or accelerometer-derived surrogates, such as tibial shock, with peak or impulse tibial bone forces during nonlevel running may be problematic; however, Matijevich and colleagues (65) found GRF, and presumably the wearable-derived GRF surrogate of tibial shock, only related to modeled tibial bone forces during level running. When assessed across running on grades ranging from 9% downhill to 9% uphill, the relationships between GRF and modeled tibial bone forces were highly inconsistent and subject-specific (65). Indeed, tibial shock has temporal differences with peak tibial bone force (Fig. 3), so it should not be surprising that the two metrics are not strongly related. Interestingly, Matijevich and colleagues (65) reported that running speed is strongly correlated with peak tibial bone loads (r = 0.97) during level running. Since running speed is accurately measured via high quality GPS running watches (20), using running speed as a surrogate of in-field tibial bone loading may suffice, but only during level running. Indeed, reducing running speed rather than distance reduced risk of tibial stress injury in a probabilistic model (66).
Case-Study: Injury Perspective in the Use of Wearables in Runners
A 22-year-old female competitive runner presented with a 6-month history of patellofemoral pain that resulted in cessation of running. After providing consent to treatment and to serve as a case study for publication, the runner successfully completed 4 wk of progressive physical therapy using current best practices, including progressive posterolateral hip and quadriceps strengthening (67). Once she no longer experienced more than minimal pain (pain = 0–1/10 on the 11-point visual analog scale) during squatting, jumping, and stair descent, the runner was progressed to a gradual return to run program. Once the runner reached 15 min of total run time (5 × 3 min of running, interspersed with 2-min walking bouts) during a progressive return to run program, she began to complain of 2 to 3/10 knee pain on a visual analog scale. The runner was issued a GPS running watch (Garmin™ 735XT; Garmin, Olathe, KS) with the capability to provide real-time feedback on running cadence (Fig. 4). To provide bandwidth feedback (68), the GPS watch was configured to provide an auditory alarm if her run cadence fell outside a range corresponding to 5% to 10% above her preferred cadence (target range, 177 to 185 steps per minute: baseline running cadence: 168 steps per minute at 4:54/km). The cadence target was chosen due to previous reports that demonstrated an 11% to 22% reduction in peak patellofemoral contact forces when running with a 5% to 10% increase in cadence compared with preferred cadence (53,58,59). She was asked to progressively increase her run duration over the next 8 sessions while maintaining the prescribed cadence. The remaining run sessions were completed during outdoor running, which was her normal training environment. Within the first three run sessions, the runner experienced a reduction in knee pain (0–1/10 pain) with the higher run cadence. Once the runner completed her eighth gait retraining session, the bandwidth feedback was discontinued. The runner continued to progress her run distance but was encouraged to check her running watch after each run to receive post-session feedback on her cadence performance for the next month. Using this program, the runner was able to return to running with less pain, perhaps at a faster rate, with a wearable device to cue an increase in run cadence. Additionally, the runner completed all gait retraining sessions in her normal training environment, which allowed her to complete her rehabilitation on her own.
Summary Recommendations for Wearable Device Use in Runners
Based on the literature reviewed in this article, recommendations can be made in the use of wearable devices to assess training loads that may relate to performance and injury. Overall, current wearable technologies accurately quantify certain running biomechanics and external training loads experienced globally by the runner. Specifically, running distance, speed, step frequency, step counts, and surrogates of the peak vertical GRF can be provided by wearables. Even at a basic level, prescribing or monitoring training loads for optimizing performance or injury rehabilitation solely based on step count, that is, loading cycles, may provide greater insight into biomechanical loads experienced by a runner versus assessing run distance alone (69). Validated wearable devices also can provide real-time, in-field feedback on running cadence, and running speed that is comparable in accuracy to respective gold standard measures.
Based on current advances, precisely tracking in-field training loads at the tissue level, such as tibial bone or Achilles tendon forces, is not currently possible solely with the use of wearable devices. The estimation of loading exposure at the tissue level during in-field running only seems presently feasible by pairing laboratory gait data with the use of wearables to track in-field loading cycles (16,44). In this scenario, subject-specific joint and tissue loads per step are quantified across a range of running speeds in a laboratory session via musculoskeletal modeling. Computational algorithms would then estimate tissue-level loads experienced by the runner by combining the laboratory-derived metrics, that is, Achilles tendon force per step, with wearable-derived data, that is, run speed and step count per training session (16,44). Knowledge of specific tissue loads can thus inform the precise prescription of training loads but are likely only feasible in highly specialized settings, that is, elite sport, where such instrumentation is available.
Future Developments for the Use of Wearables in Runners
With the rapid growth in wearable technology, new metrics are becoming a common selling point. Yet, these metrics are not always evidence-informed and may not provide beneficial insight into workload. Based on the evidence provided above and our practical insights, we propose the following four suggestions for the developments of wearable metrics that may be informative: 1) inclusion of physical activity, 2) identification of optimal cadence/stride length, 3) inclusion of pain scores alongside workload metrics, and 4) ability to quantify anatomical site-specific load.
To develop a complete picture of external workload undertaken by a runner, physical activity data should be incorporated into workload algorithms. The current focus on training/competition workload does not consider such data. Sophisticated algorithms and further developments are required to understand the effect external load measured in this way has on performance and injury risk. Wearable companies are encouraged to engage with physical activity and sport science experts to progress this area.
It is now well known that optimal cadences and stride lengths can be determining using simple algorithms and more sophisticated ones (42), but these processes still require user input. However, advancements in heart rate monitors with integrated accelerometers that determine cadence could remove the need for user input. Ideally, this would be a programmable test, allowing the mean heart rate and cadence to be calculated during steady state running. This would allow runners to fine-tune their cadence and stride length and, potentially, fast-track their physiological and mechanical adaptation to minimize their metabolic cost (see Morgan et al (70) for an example of such an intervention).
The possibility of incorporating pain-related problems that could be tracked alongside internal and external loads, provides an exciting avenue combining both injury and performance data. This underutilized approach in wearable devices could provide clinicians, coaches, and researchers with invaluable data to develop and monitor return-to-running programs, tailored race programs, and generate prospective data to identify injury risk factors.
Currently, the tracking of training loads largely focuses on loads applied to the whole person, rather than specific anatomical sites. Ultimately, researchers should aim to combine musculoskeletal modeling techniques with wearable sensors to provide tissue-level estimates of internal loads (16). By incorporating models of a tissue’s response to load (44), either catabolic or anabolic, may provide precise guidance for the frequently injured or rehabilitating runner.
The authors declare no conflict of interest and do not have any financial disclosures.
I.M. received funding from the British Association of Sport and Exercise Science to develop the software for mathematically determining optimal running cadence.
1. Running USA. Running USA NRS 2017. [Internet]. [cited 2019 June 6]. Available from: https://cdn.trustedpartner.com/docs/library/RunningUSA2012/RunningUSA_NRS_2017.pdf
2. Liu S. Fitness & activity tracker — Statistics & Facts 2019 [Internet]. [cited 2019 June 6]. Available from: https://www.statista.com/topics/4393/fitness-and-activity-tracker/
3. Peake JM, Kerr G, Sullivan JP. A critical review of consumer wearables, mobile applications, and equipment for providing biofeedback, monitoring stress, and sleep in physically active populations. Front. Physiol
. 2018; 9:743.
4. Willy RW. Innovations and pitfalls in the use of wearable devices in the prevention and rehabilitation of running related injuries. Phys. Ther. Sport
. 2018; 29:26–33.
5. Raper DP, Witchalls J, Philips EJ, et al. Use of a tibial accelerometer to measure ground reaction force in running: a reliability and validity comparison with force plates. J. Sci. Med. Sport
. 2018; 21:84–8.
6. Milner CE, Ferber R, Pollard CD, et al. Biomechanical factors associated with tibial stress fracture in female runners. Med. Sci. Sports Exerc
. 2006; 38:323–8.
7. Ruder M, Jamison ST, Tenforde A, et al. Relationship of footstrike pattern and landing impacts during a marathon. Med. Sci. Sports Exerc
. 2019; 51:2073–9.
8. Derrick TR, Dereu D, Mclean SP. Impacts and kinematic adjustments during an exhaustive run. Med. Sci. Sports Exerc
. 2002; 34:998–1002.
9. Clansey AC, Hanlon M, Wallace ES, Lake MJ. Effects of fatigue on running mechanics associated with tibial stress fracture risk. Med. Sci. Sports Exerc
. 2012; 44:1917–23.
10. Impellizzeri FM, Marcora SM, Coutts AJ. Internal and external training load: 15 years on. Int. J. Sports Physiol. Perform
. 2019; 14:270–3.
11. Renner KE, Williams D, Queen RM. The reliability and validity of the Loadsol® under various walking and running conditions. Sensors (Basel)
. 2019; 19:265.
12. Mujika I. Quantification of training and competition loads in endurance sports: methods and applications. Int. J. Sports Physiol. Perform
. 2017; 12(Suppl. 2):S29–S217.
13. Dideriksen M, Soegaard C, Nielsen RO. Validity of self-reported running distance. J. Strength Cond. Res
. 2016; 30:1592–6.
14. Johnston R, Cahalan R, Bonnett L, et al. Training load and baseline characteristics associated with new injury/pain within an endurance sporting population: a prospective study. Int. J. Sports Physiol. Perform
. 2018; 1–28.
15. Fukashiro S, Komi PV, Järvinen M, Miyashita M. Comparison between the directly measured Achilles tendon force and the tendon force calculated from the ankle joint moment during vertical jumps. Clin. Biomech. (Bristol, Avon)
. 1993; 8:25–30.
16. Pizzolato C, Lloyd DG, Barrett RS, et al. Bioinspired technologies to connect musculoskeletal mechanobiology to the person for training and rehabilitation. Front. Comput. Neurosci
. 2017; 11:96.
17. Kiernan D, Hawkins DA, Manoukian MA, et al. Accelerometer-based prediction of running injury in National Collegiate Athletic Association track athletes. J. Biomech
. 2018; 73:201–9.
18. Rowlands AV, Stiles VH. Accelerometer counts and raw acceleration output in relation to mechanical loading. J. Biomech
. 2012; 45:448–54.
19. Stiles VH, Pearce M, Moore IS, et al. Wrist-worn accelerometry for runners: objective quantification of training load. Med. Sci. Sports Exerc
. 2018; 50:2277–84.
20. Scott MT, Scott TJ, Kelly VG. The validity and reliability of global positioning systems in team sport: a brief review. J. Strength Cond. Res
. 2016; 30:1470–90.
21. Gruber AH, Murphy SP, Vollmar JE, et al. Does non-running physical activity contribute to the risk of developing a running related overuse injury?: 3837 Board #276 June 4, 800 AM - 930 AM. Med. Sci. Sports Exerc
. 2016; 48(5S):1077.
22. Scott MT, Scott TJ, Kelly VG. The validity and reliability of global positioning systems in team sport: a brief review. J. Strength Cond. Res
. 2015; Epub 2015/10/07.
23. Bosquet L, Montpetit J, Arvisais D, Mujika I. Effects of tapering on performance: a meta-analysis. Med. Sci. Sports Exerc
. 2007; 39:1358–65.
24. Spilsbury KL, Fudge BW, Ingham SA, et al. Tapering strategies in elite British endurance runners. Eur. J. Sport Sci
. 2015; 15:367–73.
25. Boey H, Aeles J, Schütte K, Vanwanseele B. The effect of three surface conditions, speed and running experience on vertical acceleration of the tibia during running. Sports Biomech
. 2017; 16:166–76.
26. Willy R, Meardon S, Schmidt A, et al. Changes in tibiofemoral contact forces during running in response to in-field gait retraining. J. Sports Sci
. 2016; 34:1602–11.
27. Adams D, Pozzi F, Carroll A, et al. Validity and utility of a commercial GPS watch for measuring running dynamics. J. Orthop. Sports Phys. Ther
. 2016; 46:A29.
28. Moore IS. Is there an economical running technique? A review of modifiable biomechanical factors affecting running economy. Sports Med
. 2016; 46:793–807.
29. Moore IS, Jones AM, Dixon SJ. Mechanisms for improved running economy in beginner runners. Med. Sci. Sports Exerc
. 2012; 44:1756–63.
30. Moore IS, Jones AM, Dixon SJ. Reduced oxygen cost of running is related to alignment of the resultant GRF and leg axis vector: a pilot study. Scand. J. Med. Sci. Sports
. 2016; 26:809–15.
31. Cavanagh PR, Williams KR. The effect of stride length variation on oxygen uptake during distance running. Med. Sci. Sports Exerc
. 1982; 14:30–5.
32. de Ruiter CJ, Verdijk PW, Werker W, et al. Stride frequency in relation to oxygen consumption in experienced and novice runners. Eur. J. Sport Sci
. 2013; 14:251–8.
33. Hunter I, Smith GA. Preferred and optimal stride frequency, stiffness and economy: changes with fatigue during a 1-h high-intensity run. Eur. J. Appl. Physiol
. 2007; 100:653–61.
34. Dorschky E, Nitschke M, Seifer AK, et al. Estimation of gait kinematics and kinetics from inertial sensor data using optimal control of musculoskeletal models. J. Biomech
. 2019; 95:109278.
35. Koska D, Gaudel J, Hein T, Maiwald C. Validation of an inertial measurement unit for the quantification of rearfoot kinematics during running. Gait Posture
. 2018; 64:135–40.
36. Wouda FJ, Giuberti M, Bellusci G, et al. Estimation of vertical ground reaction forces and sagittal knee kinematics during running using three inertial sensors. Front. Physiol
. 2018; 9:218.
37. Neal BS, Barton CJ, Gallie R, et al. Runners with patellofemoral pain have altered biomechanics which targeted interventions can modify: a systematic review and meta-analysis. Gait Posture
. 2016; 45:69–82.
38. Cheung RT, An WW, Au IP, et al. Measurement agreement between a newly developed sensing insole and traditional laboratory-based method for footstrike pattern detection in runners. PLoS One
. 2017; 12:e0175724.
39. Gruber AH, Umberger BR, Braun B, Hamill J. Economy and rate of carbohydrate oxidation during running with rearfoot and forefoot strike patterns. J. Appl. Physiol. (1985)
. 2013; 115:194–201.
40. Morris JB, Goss DL, Miller EM, Davis IS. Using real-time biofeedback to alter running biomechanics: a randomized controlled trial. Translational Sports Med
41. Salo AIT, Bezodis IN, Batterham AM, Kerwin DG. Elite sprinting: are athletes individually step-frequency or step-length reliant? Med. Sci. Sports Exerc
. 2011; 43:1055–62.
42. Moore I. Software to determine the modeled optimal gait characteristic (runtime needed). Figshare
43. Bertelsen ML, Hulme A, Petersen J, et al. A framework for the etiology of running-related injuries. Scand. J. Med. Sci. Sports
. 2017; 27:1170–80.
44. Edwards WB. Modeling overuse injuries in sport as a mechanical fatigue phenomenon. Exerc. Sport Sci. Rev
. 2018; 46:224–31.
45. Ramskov D, Rasmussen S, Sørensen H, et al. Run clever — no difference in risk of injury when comparing progression in running volume and running intensity in recreational runners: a randomised trial. BMJ Open Sport Exerc. Med
. 2018; 4:e000333.
46. Nielsen RØ, Bertelsen ML, Ramskov D, et al. The Garmin-RUNSAFE Running Health Study on the aetiology of running-related injuries: rationale and design of an 18-month prospective cohort study including runners worldwide. BMJ Open
. 2019; 9:e032627.
47. Cavanagh PR, Kram R. Stride length in distance running: velocity, body dimensions, and added mass effects. Med. Sci. Sports Exerc
. 1989; 21:467–79.
48. Edwards WB, Taylor D, Rudolphi TJ, et al. Effects of stride length and running mileage on a probabilistic stress fracture model. Med. Sci. Sports Exerc
. 2009; 41:2177–84.
49. Heiderscheit BC, Chumanov ES, Michalski MP, et al. Effects of step rate manipulation on joint mechanics during running. Med. Sci. Sports Exerc
. 2011; 43:296–302.
50. Luedke LE, Heiderscheit BC, Williams DS, Rauh MJ. Influence of step rate on shin injury and anterior knee pain in high school runners. Med. Sci. Sports Exerc
. 2016; 48:1244–50.
51. Wille C, Lenhart RL, Wang S, et al. Ability of sagittal kinematic variables to estimate ground reaction forces and joint kinetics in running. J. Orthop. Sports Phys. Ther
. 2014; 44:825–30.
52. Napier C, MacLean CL, Maurer J, et al. Kinematic correlates of kinetic outcomes associated with running-related injury. J. Appl. Biomech
. 2019; 35:123–30.
53. Lenhart RL, Thelen DG, Wille CM, et al. Increasing running step rate reduces patellofemoral joint forces. Med. Sci. Sports Exerc
. 2013; 46:557–64.
54. Bowersock CD, Willy RW, DeVita P, Willson JD. Reduced step length reduces knee joint contact forces during running following anterior cruciate ligament reconstruction but does not alter inter-limb asymmetry. Clin. Biomech. (Bristol, Avon)
. 2017; 43:79–85.
55. Willy RW, Buchenic L, Rogacki K, et al. In-field gait retraining and mobile monitoring to address running biomechanics associated with tibial stress fracture. Scand. J. Med. Sci. Sports
. 2016; 26:197–205.
56. Firminger CR, Fung A, Loundagin LL, Edwards WB. Effects of footwear and stride length on metatarsal strains and failure in running. Clin. Biomech. (Bristol, Avon)
. 2017; 49:8–15.
57. Wellenkotter J, Kernozek TW, Meardon S, Suchomel T. The effects of running cadence manipulation on plantar loading in healthy runners. Int. J. Sports Med
. 2014; 35:779–84.
58. Willson JD, Sharpee R, Meardon SA, Kernozek TW. Effects of step length on patellofemoral joint stress in female runners with and without patellofemoral pain. Clin. Biomech. (Bristol, Avon)
. 2014; 29:243–7.
59. Willy RW, Willson JD, Clowers K, et al. The effects of body-borne loads and cadence manipulation on patellofemoral and tibiofemoral joint kinetics during running. J. Biomech
. 2016; 49:4028–33.
60. Edwards WB, Ward ED, Meardon SA, Derrick TR. The use of external transducers for estimating bone strain at the distal tibia during impact activity. J. Biomech. Eng
. 2009; 131:51009.
61. Cheung RTH, Zhang JH, Chan ZYS, et al. Shoe-mounted accelerometers should be used with caution in gait retraining. Scand. J. Med. Sci. Sports
. 2019; 29:835–42.
62. Crowell HP, Davis IS. Gait retraining to reduce lower extremity loading in runners. Clin. Biomech. (Bristol, Avon)
. 2011; 26:78–83.
63. Beaupré G, Orr T, Carter D. An approach for time-dependent bone modeling and remodeling—theoretical development. J. Orthop. Res
. 1990; 8:651–61.
64. Besier T. The importance of measuring lower limb cumulative load in sport: a mechanobiological approach: white paper. 2018 [cited 2019 June 23]. Available from: https://imeasureu.com/2018/02/26/measuring-lower-limb-cumulative-load-sport/
65. Matijevich ES, Branscombe LM, Scott LR, Zelik KE. Ground reaction force metrics are not strongly correlated with tibial bone load when running across speeds and slopes: Implications for science, sport and wearable tech. PLoS One
. 2019; 14:e0210000.
66. Edwards WB, Taylor D, Rudolphi TJ, et al. Effects of running speed on a probabilistic stress fracture model. Clin. Biomech. (Bristol, Avon)
. 2010; 25:372–7.
67. Willy RW, Hoglund LT, Barton CJ, et al. Patellofemoral pain: Clinical practice guidelines linked to the international classification of functioning, disability and health from the Academy of Orthopaedic Physical Therapy of the American Physical Therapy Association. J. Orthop. Sports Phys. Ther
. 2019; 49:CPG1–95.
68. Whittier T, Willy RW, Sandri Heidner G, et al. The cognitive demands of gait retraining in runners: an EEG study. J. Mot. Behav
. 2019; 1–12.
69. Stellingwerff T. Case study: Body composition periodization in an Olympic-level female middle-distance runner over a 9-year career. Int. J. Sport Nutr. Exerc. Metab
. 2018; 28:428–33.
70. Morgan D, Martin P, Craib M, et al. Effect of step length optimization on the aerobic demand of running. J. Appl. Physiol
. 1994; 77.