Using wearable sensors to classify subject-specific running biomechanical gait patterns based on changes in environmental weather conditions

Longitudinal Monitoring of Biomechanical and Psychological State in Collegiate Female Basketball Athletes Using Principal Component Analysis

Background

The growth in participation in collegiate athletics has been accompanied by increased sport-related injuries. The complex and multifactorial nature of sports injuries highlights the importance of monitoring athletes prospectively using a novel and integrated biopsychosocial approach, as opposed to contemporary practices that silo these facets of health.

Methods

Data collected over two competitive basketball seasons were used in a principal component analysis (PCA) model with the following objectives: (i) investigate whether biomechanical PCs (i.e., on-court and countermovement jump (CMJ) metrics) were correlated with psychological state across a season and (ii) explore whether subject-specific significant fluctuations could be detected using minimum detectable change statistics. Weekly CMJ (force plates) and on-court data (inertial measurement units), as well as psychological state (questionnaire) data, were collected on the female collegiate basketball team for two seasons.

Results

While some relationships (n = 2) were identified between biomechanical PCs and psychological state metrics, the magnitude of these associations was weak (r = |0.18-0.19|, p < 0.05), and no other overarching associations were identified at the group level. However, post-hoc case study analysis showed subject-specific relationships that highlight the potential utility of red-flagging meaningful fluctuations from normative biomechanical and psychological patterns.

Conclusion

Overall, this work demonstrates the potential of advanced analytical modeling to characterize components of and detect statistically and clinically relevant fluctuations in student-athlete performance, health, and well-being and the need for more tailored and athlete-centered monitoring practices.

Relationship Between Running Biomechanics and Core Temperature Across a Competitive Road Race

BACKGROUND

Outdoor races introduce environmental stressors to runners, and core temperature changes may influence runners' movement patterns. This study assessed changes and determined relationships between sensor-derived running biomechanics and core temperature among runners across an 11.27-km road race.

HYPOTHESIS

Core temperatures would increase significantly across the race, related to changes in spatiotemporal biomechanical measures.

STUDY DESIGN

Cross-sectional cohort study.

LEVEL OF EVIDENCE

Level 3.

METHODS

Twenty runners (9 female, 11 male; age, 48 ± 12 years; height, 169.7 ± 9.1 cm; mass, 71.3 ± 13.4 kg) enrolled in the 2022 Falmouth Road Race were recruited. Participants used lightweight technologies (ingestible thermistors and wearable sensors) to monitor core temperature and running biomechanics throughout the race. Timestamps were used to align sensor-derived measures for 7 race segments. Observations were labeled as core temperatures generally within normal limits (<38°C) or at elevated core temperatures (≥38°C). Multivariate repeated measures analyses of variance were used to assess changes in sensor-derived measures across the race, with Bonferroni post hoc comparisons for significant findings. Pearson's r correlations were used to assess the relationship between running biomechanics and core temperature measures.

RESULTS

Eighteen participants developed hyperthermic core temperatures (39.0°C ± 0.5°C); core temperatures increased significantly across the race (P < 0.01). Kinetic measures obtained from the accelerometers, including shock, impact, and braking g, all significantly increased across the race (P < 0.01); other sensor-derived biomechanical measures did not change significantly. Core temperatures were weakly associated with biomechanics (|r range|, 0.02-0.16).

CONCLUSION

Core temperatures and kinetics increased significantly across a race, yet these outcomes were not strongly correlated. The observed kinetic changes may have been attributed to fatigue-related influences over the race.

CLINICAL RELEVANCE

Clinicians may not expect changes in biomechanical movement patterns to signal thermal responses during outdoor running in a singular event.

Reliability of Xsens IMU-Based Lower Extremity Joint Angles during In-Field Running

The Xsens Link motion capture suit has become a popular tool in investigating 3D running kinematics based on wearable inertial measurement units outside of the laboratory. In this study, we investigated the reliability of Xsens-based lower extremity joint angles during unconstrained running on stable (asphalt) and unstable (woodchip) surfaces within and between five different testing days in a group of 17 recreational runners (8 female, 9 male). Specifically, we determined the within-day and between-day intraclass correlation coefficients (ICCs) and minimal detectable changes (MDCs) with respect to discrete ankle, knee, and hip joint angles. When comparing runs within the same day, the investigated Xsens-based joint angles generally showed good to excellent reliability (median ICCs > 0.9). Between-day reliability was generally lower than the within-day estimates: Initial hip, knee, and ankle angles in the sagittal plane showed good reliability (median ICCs > 0.88), while ankle and hip angles in the frontal plane showed only poor to moderate reliability (median ICCs 0.38-0.83). The results were largely unaffected by the surface. In conclusion, within-day adaptations in lower-extremity running kinematics can be captured with the Xsens Link system. Our data on between-day reliability suggest caution when trying to capture longitudinal adaptations, specifically for ankle and hip joint angles in the frontal plane.

Subject-specific sensitivity of several biomechanical features to fatigue during an exhaustive treadmill run

The aim of the present study was to examine the sensitivity of several movement features during running to exhaustion in a subject-specific setup adopting a cross-sectional design and a machine learning approach. Thirteen recreational runners, that systematically trained and competed, performed an exhaustive running protocol on an instrumented treadmill. Respiratory data were collected to establish the second ventilatory threshold (VT2) in order to obtain a reference point regarding the gradual accumulation of fatigue. A machine learning approach was adopted to analyze kinetic and kinematic data recorded for each participant, using a random forest classifier for the region pre and post the second ventilatory threshold. SHapley Additive exPlanations (SHAP) analysis was used to explain the models' predictions and to provide insight about the most important variables. The classification accuracy value of the models adopted ranged from 0.853 to 0.962. The most important feature in six out of thirteen participants was the angular range in AP axis of upper trunk C7 (RTAPu) followed by maximum loading rate (RFDmaxD) and the angular range in the LT axis of the C7. SHAP dependence plots also showed an increased dispersion of predictions in stages around the second ventilatory threshold which is consistent with feature interactions. These results showed that each runner used the examined features differently to cope with the increase in fatigue and mitigate its effects in order to maintain a proper motor pattern.

Use of subject-specific models to detect fatigue-related changes in running biomechanics: a random forest approach

Running biomechanics are affected by fatiguing or prolonged runs. However, no evidence to date has conclusively linked this effect to running-related injury (RRI) development or performance implications. Previous investigations using subject-specific models in running have demonstrated higher accuracy than group-based models, however, this has been infrequently applied to fatigue. In this study, two experiments were conducted to determine whether subject-specific models outperformed group-based models to classify running biomechanics during non-fatigued and fatigued conditions. In the first experiment, 16 participants performed four treadmill runs at or around the maximal lactate steady state. In the second experiment, nine participants performed five prolonged runs using commercial wearable devices. For each experiment, two segments were extracted from each trial from early and late in the run. For each participant, a random forest model was applied with a leave-one-run-out cross-validation to classify between the early (non-fatigued) and late (fatigued) segments. Additionally, group-based classifiers with a leave-one-subject-out cross validation were constructed. For experiment 1, mean classification accuracies for the single-subject and group-based classifiers were 68.2 ± 8.2% and 57.0 ± 8.9%, respectively. For experiment 2, mean classification accuracies for the single-subject and group-based classifiers were 68.9 ± 17.1% and 61.5 ± 11.7%, respectively. Variable importance rankings were consistent within participants, but these rankings differed from each participant to those of the group. Although the classification accuracies were relatively low, these findings highlight the advantage of subject-specific classifiers to detect changes in running biomechanics with fatigue and indicate the potential of using big data and wearable technology approaches in future research to determine possible connections between biomechanics and RRI.

Sensitiveness of Variables Extracted from a Fitness Smartwatch to Detect Changes in Vertical Impact Loading during Outdoors Running

Accelerometry is becoming a popular method to access human movement in outdoor conditions. Running smartwatches may acquire chest accelerometry through a chest strap, but little is known about whether the data from these chest straps can provide indirect access to changes in vertical impact properties that define rearfoot or forefoot strike. This study assessed whether the data from a fitness smartwatch and chest strap containing a tri-axial accelerometer (FS) is sensible to detect changes in running style. Twenty-eight participants performed 95 m running bouts at ~3 m/s in two conditions: normal running and running while actively reducing impact sounds (silent running). The FS acquired running cadence, ground contact time (GCT), stride length, trunk vertical oscillation (TVO), and heart rate. Moreover, a tri-axial accelerometer attached to the right shank provided peak vertical tibia acceleration (PKACC). The running parameters extracted from the FS and PKACC variables were compared between normal and silent running. Moreover, the association between PKACC and smartwatch running parameters was accessed using Pearson correlations. There was a 13 ± 19% reduction in PKACC (p < 0.005), and a 5 ± 10% increase in TVO from normal to silent running (p < 0.01). Moreover, there were slight reductions (~2 ± 2%) in cadence and GCT when silently running (p < 0.05). However, there were no significant associations between PKACC and the variables extracted from the FS (r < 0.1, p > 0.05). Therefore, our results suggest that biomechanical variables extracted from FS have limited sensitivity to detect changes in running technique. Moreover, the biomechanical variables from the FS cannot be associated with lower limb vertical loading.

Recent Machine Learning Progress in Lower Limb Running Biomechanics With Wearable Technology: A Systematic Review

With the emergence of wearable technology and machine learning approaches, gait monitoring in real-time is attracting interest from the sports biomechanics community. This study presents a systematic review of machine learning approaches in running biomechanics using wearable sensors. Electronic databases were retrieved in PubMed, Web of Science, SPORTDiscus, Scopus, IEEE Xplore, and ScienceDirect. A total of 4,068 articles were identified via electronic databases. Twenty-four articles that met the eligibility criteria after article screening were included in this systematic review. The range of quality scores of the included studies is from 0.78 to 1.00, with 40% of articles recruiting participant numbers between 20 and 50. The number of inertial measurement unit (IMU) placed on the lower limbs varied from 1 to 5, mainly in the pelvis, thigh, distal tibia, and foot. Deep learning algorithms occupied 57% of total machine learning approaches. Convolutional neural networks (CNN) were the most frequently used deep learning algorithm. However, the validation process for machine learning models was lacking in some studies and should be given more attention in future research. The deep learning model combining multiple CNN and recurrent neural networks (RNN) was observed to extract different running features from the wearable sensors and presents a growing trend in running biomechanics.

Is This the Real Life, or Is This Just Laboratory? A Scoping Review of IMU-Based Running Gait Analysis

Inertial measurement units (IMUs) can be used to monitor running biomechanics in real-world settings, but IMUs are often used within a laboratory. The purpose of this scoping review was to describe how IMUs are used to record running biomechanics in both laboratory and real-world conditions. We included peer-reviewed journal articles that used IMUs to assess gait quality during running. We extracted data on running conditions (indoor/outdoor, surface, speed, and distance), device type and location, metrics, participants, and purpose and study design. A total of 231 studies were included. Most (72%) studies were conducted indoors; and in 67% of all studies, the analyzed distance was only one step or stride or <200 m. The most common device type and location combination was a triaxial accelerometer on the shank (18% of device and location combinations). The most common analyzed metric was vertical/axial magnitude, which was reported in 64% of all studies. Most studies (56%) included recreational runners. For the past 20 years, studies using IMUs to record running biomechanics have mainly been conducted indoors, on a treadmill, at prescribed speeds, and over small distances. We suggest that future studies should move out of the lab to less controlled and more real-world environments.

Knowledge and management of low back pain as running-related injuries among Italian physical therapists: findings from a national survey

OBJECTIVES

To investigate the beliefs, knowledge, attitudes, behavior, and the clinical management procedures of the Italian physical therapists specialized in orthopedic manipulative physical therapy (OMPT) toward running and its correlation with low back pain (LBP).Design: A cross-sectional online survey was conducted in 2019, according to the Checklist for Reporting Results of Internet E-Surveys (CHERRIES) and Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.Setting: Italy.Participants: One thousand two hundred and eighteen Italian OMPTs.

METHODS

Survey Monkey software was used to administer the survey. The questionnaire was self-reported and included 26 questions. Descriptive statistics were used and related to the effective respondents for each question.

RESULTS

One thousand two hundred and eighteen questionnaires (60.9%) were included in the analysis. A considerable cohort of OMPTs working in private practice clinical settings (n = 845; 69.4%; 95% CI 66.7-71.9) has indicated running not to be a relevant risk factor for the onset of LBP (n = 806; 66.2%; 95% CI 63.4-68.8). Moreover, most of the participants (n = 679; 55.7%; 95% CI 52.9-58.5) adopted a combination of manual therapy techniques and therapeutic exercise for the management of runners with LBP.

CONCLUSIONS

Widespread knowledge of clinical and theoretical management of LBP in runners-patients has emerged among Italian OMPTs. The OMPTs' academic background agrees with the recent literature and therefore highlights the paucity of studies related to LBP as running-related injuries.

Using a Deep Learning Method and Data from Two-Dimensional (2D) Marker-Less Video-Based Images for Walking Speed Classification

Human body measurement data related to walking can characterize functional movement and thereby become an important tool for health assessment. Single-camera-captured two-dimensional (2D) image sequences of marker-less walking individuals might be a simple approach for estimating human body measurement data which could be used in walking speed-related health assessment. Conventional body measurement data of 2D images are dependent on body-worn garments (used as segmental markers) and are susceptible to changes in the distance between the participant and camera in indoor and outdoor settings. In this study, we propose five ratio-based body measurement data that can be extracted from 2D images and can be used to classify three walking speeds (i.e., slow, normal, and fast) using a deep learning-based bidirectional long short-term memory classification model. The results showed that average classification accuracies of 88.08% and 79.18% could be achieved in indoor and outdoor environments, respectively. Additionally, the proposed ratio-based body measurement data are independent of body-worn garments and not susceptible to changes in the distance between the walking individual and camera. As a simple but efficient technique, the proposed walking speed classification has great potential to be employed in clinics and aged care homes.

The Relationship between VO2max, Power Management, and Increased Running Speed: Towards Gait Pattern Recognition through Clustering Analysis

Triathlon has become increasingly popular in recent years. In this discipline, maximum oxygen consumption (VO₂max) is considered the gold standard for determining competition cardiovascular capacity. However, the emergence of wearable sensors (as Stryd) has drastically changed training and races, allowing for the more precise evaluation of athletes and study of many more potential determining variables. Thus, in order to discover factors associated with improved running efficiency, we studied which variables are correlated with increased speed. We then developed a methodology to identify associated running patterns that could allow each individual athlete to improve their performance. To achieve this, we developed a correlation matrix, implemented regression models, and created a heat map using hierarchical cluster analysis. This highlighted relationships between running patterns in groups of young triathlon athletes and several different variables. Among the most important conclusions, we found that high VO₂max did not seem to be significantly correlated with faster speed. However, faster individuals did have higher power per kg, horizontal power, stride length, and running effectiveness, and lower ground contact time and form power ratio. VO₂max appeared to strongly correlate with power per kg and this seemed to indicate that to run faster, athletes must also correctly manage their power.

Usability of Wearable Devices With a Novel Cardiac Force Index for Estimating the Dynamic Cardiac Function: Observational Study

Background

Long-distance running can be a form of stress to the heart. Technological improvements combined with the public’s gradual turn toward mobile health (mHealth), self-health, and exercise effectiveness have resulted in the widespread use of wearable exercise products. The monitoring of dynamic cardiac function changes during running and running performance should be further studied.

Objective

We investigated the relationship between dynamic cardiac function changes and finish time for 3000-meter runs. Using a wearable device based on a novel cardiac force index (CFI), we explored potential correlations among 3000-meter runners with stronger and weaker cardiac functions during running.

Methods

This study used the American product BioHarness 3.0 (Zephyr Technology Corporation), which can measure basic physiological parameters including heart rate, respiratory rate, temperature, maximum oxygen consumption, and activity. We investigated the correlations among new physiological parameters, including CFI = weight * activity / heart rate, cardiac force ratio (CFR) = CFI of running / CFI of walking, and finish times for 3000-meter runs.

Results

The results showed that waist circumference, smoking, and CFI were the significant factors for qualifying in the 3000-meter run. The prediction model was as follows: ln (3000 meters running performance pass probability / fail results probability) = –2.702 – 0.096 × [waist circumference] – 1.827 × [smoke] + 0.020 × [ACi7]. If smoking and the ACi7 were controlled, contestants with a larger waist circumference tended to fail the qualification based on the formula above. If waist circumference and ACi7 were controlled, smokers tended to fail more often than nonsmokers. Finally, we investigated a new calculation method for monitoring cardiac status during exercise that uses the CFI of walking for the runner as a reference to obtain the ratio between the cardiac force of exercise and that of walking (CFR) to provide a standard for determining if the heart is capable of exercise. A relationship is documented between the CFR and the performance of 3000-meter runs in a healthy 22-year-old person. During the running period, data are obtained while participant slowly runs 3000 meters, and the relationship between the CFR and time is plotted. The runner’s CFR varies with changes in activity. Since the runner’s acceleration increases, the CFR quickly increases to an explosive peak, indicating the runner’s explosive power. At this period, the CFI revealed a 3-fold increase (CFR=3) in a strong heart. After a time lapse, the CFR is approximately 2.5 during an endurance period until finishing the 3000-meter run. Similar correlation is found in a runner with a weak heart, with the CFR at the beginning period being 4 and approximately 2.5 thereafter.

Conclusions

In conclusion, the study results suggested that measuring the real-time CFR changes could be used in a prediction model for 3000-meter running performance.

Design of a Low-Cost, Wearable Device for Kinematic Analysis in Physical Therapy Settings

BACKGROUND

Unsupervised home exercise is a major component of physical therapy (PT). This study proposes an inexpensive, inertial measurement unit-based wearable device to capture kinematic data to facilitate exercise. However, conveying and interpreting kinematic data to non-experts poses a challenge due to the complexity and background knowledge required that most patients lack.

OBJECTIVES

The objectives of this study were to identify key user interface and user experience features that would likely improve device adoption and assess participant receptiveness toward the device.

METHODS

Fifty participants were recruited to perform nine upper extremity exercises while wearing the device. Prior to exercise, participants completed an orientation of the device, which included examples of software graphics with exercise data. Surveys that measured receptiveness toward the device, software graphics, and ergonomics were given before and after exercise.

RESULTS

Participants were highly receptive to the device with 90% of the participants likely to use the device during PT. Participants understood how the simple kinematic data could be used to aid exercise, but the data could be difficult to comprehend with more complex movements. Devices should incorporate wireless sensors and emphasize ease of wear.

CONCLUSION

Device-guided home physical rehabilitation can allow for individualized treatment protocols and improve exercise self-efficacy through kinematic analysis. Future studies should implement clinical testing to evaluate the impact a wearable device can have on rehabilitation outcomes.

Wearable Sensor-Based Gait Analysis for Age and Gender Estimation

Wearable sensor-based systems and devices have been expanded in different application domains, especially in the healthcare arena. Automatic age and gender estimation has several important applications. Gait has been demonstrated as a profound motion cue for various applications. A gait-based age and gender estimation challenge was launched in the 12th IAPR International Conference on Biometrics (ICB), 2019. In this competition, 18 teams initially registered from 14 countries. The goal of this challenge was to find some smart approaches to deal with age and gender estimation from sensor-based gait data. For this purpose, we employed a large wearable sensor-based gait dataset, which has 745 subjects (357 females and 388 males), from 2 to 78 years old in the training dataset; and 58 subjects (19 females and 39 males) in the test dataset. It has several walking patterns. The gait data sequences were collected from three IMUZ sensors, which were placed on waist-belt or at the top of a backpack. There were 67 solutions from ten teams-for age and gender estimation. This paper extensively analyzes the methods and achieved-results from various approaches. Based on analysis, we found that deep learning-based solutions lead the competitions compared with conventional handcrafted methods. We found that the best result achieved 24.23% prediction error for gender estimation, and 5.39 mean absolute error for age estimation by employing angle embedded gait dynamic image and temporal convolution network.

New Considerations for Collecting Biomechanical Data Using Wearable Sensors: The Effect of Different Running Environments

Traditionally, running biomechanics analyses have been conducted using 3D motion capture during treadmill or indoor overground running. However, most runners complete their runs outdoors. Since changes in running terrain have been shown to influence running gait mechanics, the purpose of this study was to use a machine learning approach to objectively determine relevant accelerometer-based features to discriminate between running patterns in different environments and determine the generalizability of observed differences in running patterns. Center of mass accelerations were recorded for recreational runners in treadmill-only (n = 28) and sidewalk-only (n = 25) environments, and an independent group (n = 16) ran in both treadmill and sidewalk environments. A feature selection algorithm was used to develop a training dataset from treadmill-only and sidewalk-only running. A binary support vector machine model was trained to classify treadmill and sidewalk running. Classification accuracy was determined using 10-fold cross-validation of the training dataset and an independent testing dataset from the runners that ran in both environments. Nine features related to the consistency and variability of center of mass accelerations were selected. Specifically, there was greater ratio of vertical acceleration during treadmill running and a greater ratio of anterior-posterior acceleration during sidewalk running in both the training and testing dataset. Step and stride regularity were significantly greater in the treadmill condition for the vertical axis in both the training and testing dataset, and in the medial-lateral axis for the testing dataset. During sidewalk running, there was significantly greater variability in the magnitude of the vertical and anterior-posterior accelerations for both datasets. The classification accuracy based on 10-fold cross-validation of the training dataset (M = 93.17%, SD = 2.43%) was greater than the classification accuracy of the independent testing dataset (M = 83.81%, SD = 3.39%). This approach could be utilized in future analyses to identify relevant differences in running patterns using wearable technology.

Measuring markers of aging and knee osteoarthritis gait using inertial measurement units

Differences in gait with age or knee osteoarthritis have been demonstrated in laboratory studies using optical motion capture (MoCap). While MoCap is accurate and reliable, it is impractical for assessment outside the laboratory. Inertial measurement units (IMUs) may be useful in these situations. Before IMUs are used as a surrogate for MoCap, methods that are reliable, repeatable, and that calculate metrics at similar accuracy to MoCap must be demonstrated. The purpose of this study was to compare spatiotemporal gait parameters and knee range of motion calculated via MoCap to IMU-derived variables and to compare the ability of these tools to discriminate between groups. MoCap and IMU data were collected from young, older, and adults with knee osteoarthritis during overground walking at three self-selected speeds. Walking velocity, stride length, cadence, percent of gait cycle in stance, and sagittal knee range of motion were calculated and compared between tools (MoCap and IMU), between participant groups, and across speed. There were no significant differences between MoCap and IMU outcomes, and root mean square error between tools was ≤0.05 m/s for walking velocity, ≤0.07 m for stride length, ≤0.5 strides/min for cadence, ≤5% for percent of gait cycle in stance, and ≤1.5° for knee range of motion. No interactions were present, suggesting that MoCap and IMU calculated metrics similarly across groups and speeds. These results demonstrate IMUs can accurately calculate spatiotemporal variables and knee range of motion during gait in young and older, asymptomatic and knee osteoarthritis cohorts.

New Considerations for Wearable Technology Data: Changes in Running Biomechanics During a Marathon

The purpose of this study was to use wearable technology data to quantify alterations in subject-specific running patterns throughout a marathon race and to determine if runners could be clustered into subgroups based on similar trends in running gait alterations throughout the marathon. Using a wearable sensor, data were collected for cadence, braking, bounce, pelvic rotation, pelvic drop, and ground contact time for 27 runners. A composite index was calculated based on the "typical" data (4-14 km) for each runner and evaluated for 14 individual 2-km sections thereafter to detect "atypical" data (ie, higher indices). A cluster analysis assigned all runners to a subgroup based on similar trends in running alterations. Results indicated that the indices became significantly higher starting at 20 to 22 km. Cluster 1 exhibited lower indices than cluster 2 throughout the marathon, and the only significant difference in characteristics between clusters was that cluster 1 had a lower age-grade performance score than cluster 2. In summary, this study presented a novel method to investigate the effects of fatigue on running biomechanics using wearable technology in a real-world setting. Recreational runners with higher age-grade performance scores had less atypical running patterns throughout the marathon compared with runners with lower age-grade performance scores.

New Considerations for Collecting Biomechanical Data Using Wearable Sensors: How Does Inclination Influence the Number of Runs Needed to Determine a Stable Running Gait Pattern?

As inertial measurement units (IMUs) are used to capture gait data in real-world environments, guidelines are required in order to determine a ‘typical’ or ‘stable’ gait pattern across multiple days of data collection. Since uphill and downhill running can greatly affect the biomechanics of running gait, this study sought to determine the number of runs needed to establish a stable running pattern during level, downhill, and uphill conditions for both univariate and multivariate analyses of running biomechanical data collected using a single wearable IMU device. Pelvic drop, ground contact time, braking, vertical oscillation, pelvic rotation, and cadence, were recorded from thirty-five recreational runners running in three elevation conditions: level, downhill, and uphill. Univariate and multivariate normal distributions were estimated from differing numbers of runs and stability was defined when the addition of a new run resulted in less than a 5% change in the 2.5 and 97.5 quantiles of the 95% probability density function for each individual runner. This stability point was determined separately for each runner and each IMU variable (univariate and multivariate). The results showed that 2–4 runs were needed to define a stable running pattern for univariate, and 4–5 days were necessary for multivariate analysis across all inclination conditions. Pearson’s correlation coefficients were calculated to cross-validate differing elevation conditions and showed excellent correlations (r = 0.98 to 1.0) comparing the training and testing data within the same elevation condition and good to very good correlations (r = 0.63–0.88) when comparing training and testing data from differing elevation conditions. These results suggest that future research involving wearable technology should collect multiple days of data in order to build reliable and accurate representations of an individual’s stable gait pattern.

Automated Accelerometer-Based Gait Event Detection During Multiple Running Conditions

The identification of the initial contact (IC) and toe off (TO) events are crucial components of running gait analyses. To evaluate running gait in real-world settings, robust gait event detection algorithms that are based on signals from wearable sensors are needed. In this study, algorithms for identifying gait events were developed for accelerometers that were placed on the foot and low back and validated against a gold standard force plate gait event detection method. These algorithms were automated to enable the processing of large quantities of data by accommodating variability in running patterns. An evaluation of the accuracy of the algorithms was done by comparing the magnitude and variability of the difference between the back and foot methods in different running conditions, including different speeds, foot strike patterns, and outdoor running surfaces. The results show the magnitude and variability of the back-foot difference was consistent across running conditions, suggesting that the gait event detection algorithms can be used in a variety of settings. As wearable technology allows for running gait analyses to move outside of the laboratory, the use of automated accelerometer-based gait event detection methods may be helpful in the real-time evaluation of running patterns in real world conditions.

Subject-specific and group-based running pattern classification using a single wearable sensor

The objective of this study was to determine whether subject-specific or group-based models provided better classification accuracy to identify changes in biomechanical running gait patterns across different inclination conditions. The classification process was based on measurements from a single wearable sensor using a total of 41,780 strides from eleven recreational runners while running in real-world and uncontrolled environment. Biomechanical variables included pelvic drop, ground contact time, braking, vertical oscillation of pelvis, pelvic rotation, and cadence were recorded during running on three inclination grades: downhill, -2° to -7°; level, -0.2° to +0.2°; and uphill, +2° to +7°. An ensemble and non-linear machine learning algorithm, random forest (RF), was used to classify inclination condition and determine the importance of each of the biomechanical variables. Classification accuracy was determined for subject-specific and group-based RF models. The mean classification accuracy of all subject-specific RF models was 86.29%, while group-based classification accuracy was 76.17%. Braking was identified as the most important variable for all the runners using the group-based model and for most of the runners based on a subject-specific models. In addition, individual runners used different strategies across different inclination conditions and the ranked order of variable importance was unique for each runner. These results demonstrate that subject-specific models can better characterize changes in gait biomechanical patterns compared to a more traditional group-based approach.