Landmark registering waveform data improves the ability to predict performance measures

Dynamic assessment for low back-support exoskeletons during manual handling tasks

Exoskeletons can protect users' lumbar spine and reduce the risk of low back injury during manual lifting tasks. Although many exoskeletons have been developed, their adoptability is limited by their task- and movement-specific effects on reducing burden. Many studies have evaluated the safety and effectiveness of an exoskeleton using the peak/mean values of biomechanical variables, whereas the performance of the exoskeleton at other time points of the movement has not been investigated in detail. A functional analysis, which presents discrete time-series data as continuous functions, makes it possible to highlight the features of the movement waveform and determine the difference in each variable at each time point. This study investigated an assessment method for exoskeletons based on functional ANOVA, which made it possible to quantify the differences in the biomechanical variables throughout the movement when using an exoskeleton. Additionally, we developed a method based on the interpolation technique to estimate the assistive torque of an exoskeleton. Ten men lifted a 10-kg box under symmetric and asymmetric conditions five times each. Lumbar load was significantly reduced during all phases (flexion, lifting, and laying) under both conditions. Additionally, reductions in kinematic variables were observed, indicating the exoskeleton's impact on motion restrictions. Moreover, the overlap F-ratio curves of the lumbar load and kinematic variables imply that exoskeletons reduce the lumbar load by restricting the kinematic variables. The results suggested that at smaller trunk angles (<25°), an exoskeleton neither significantly reduces the lumbar load nor restricts trunk movement. Our findings will help increasing exoskeleton safety and designing effective products for reducing lumbar injury risks.

The effects of curve registration on linear models of jump performance and classification based on vertical ground reaction forces

Functional principal components define modes of variation in time series, which represent characteristic movement patterns in biomechanical data. Their usefulness however depends on the prior choices made in data processing. Recent research showed that better curve alignment achieved with registration (dynamic time warping) reduces errors in linear models predicting jump height. However, the efficacy of registration in different preprocessing combinations, including time normalisation, padding and feature extraction, is largely unknown. A more comprehensive analysis is needed, given the potential value of registration to machine learning in biomechanics. We evaluated popular preprocessing methods combined with registration, creating 512 models based on ground reaction force data from 385 countermovement jumps. The models either predicted jump height or classified jumps into those performed with or without arm swing. Our results show that the classification models benefited from registration in various forms, particularly when landmarks were placed at critical points. The best classifier achieved a 5.5 percentage point improvement over the equivalent unregistered model. However, registration was detrimental to the jump height models, although this performance variable may be a special case given its direct relationship with impulse. Our meta-models revealed the relative contributions made by various preprocessing operations, highlighting that registration does not generalise so well to new data. Nonetheless, our analysis shows the potential for registration in further biomechanical applications, particularly in classification, when combined with the other appropriate preprocessing operations.

Biomechanical but Not Strength or Performance Measures Differentiate Male Athletes Who Experience ACL Reinjury on Return to Level 1 Sports

BACKGROUND

Performance measures such as strength, jump height/length, and change of direction (CoD) time during anterior cruciate ligament (ACL) rehabilitation have been used to determine readiness to return to play and identify those who may be at risk of rerupture. However, athletes may reach these criteria despite ongoing biomechanical deficits when performing these tests. Combining return-to-play criteria with an assessment of movement through 3-dimensional (3D) biomechanics in male field sports athletes to identify risk factors for ACL rerupture has not been explored previously.

PURPOSE

To prospectively examine differences in strength, jump, and CoD performance and movement using 3D biomechanics in a cohort of male athletes playing level 1 sports (ie, multidirectional field sports that involve landing, pivoting, or CoD) between those who reinjured the reconstructed ACL (RI group) and those with no reinjury (NRI group) after 2 years of follow-up and to examine the ability of these differences to predict reinjury.

STUDY DESIGN

Cohort study; Level of evidence, 2.

METHODS

After primary ACL reconstruction (ACLR), 1045 male athletes were recruited and underwent testing 9 months after surgery including isokinetic strength, jump, and CoD performance measures as well as patient-reported outcomes and 3D biomechanical analyses. Participants were followed up after 2 years regarding ACL reinjury status. Differences were determined between the RI and NRI groups in patient-reported outcomes, performance measures, and 3D biomechanics on the ACLR side and symmetry between limbs. The ability of these measures to predict ACL reinjury was determined through logistic regression.

RESULTS

No differences were identified in strength and performance measures on the ACLR side or in symmetry. Biomechanical analysis indicated differences on the ACLR side primarily in the sagittal plane for the double-leg drop jump (effect size, 0.59-0.64) and greater asymmetry primarily in the frontal plane during unplanned CoD (effect size, 0.61-0.69) in the RI group. While these biomechanical test results were different between groups, multivariate regression modeling demonstrated limited ability (area under the curve, 0.67 and 0.75, respectively) to prospectively predict ACL reinjury.

CONCLUSION

Commonly reported return-to-play strength, jump, and timed CoD performance measures did not differ between the RI and NRI groups. Differences in movement based on biomechanical measures during double-leg drop jump and unplanned CoD were identified, although they had limited ability to predict reinjury. Targeting these variables during rehabilitation may reduce reinjury risk in male athletes returning to level 1 sports after ACLR.

REGISTRATION

NCT02771548 (ClinicalTrials.gov identifier).

Can Biomechanical Testing After Anterior Cruciate Ligament Reconstruction Identify Athletes at Risk for Subsequent ACL Injury to the Contralateral Uninjured Limb?

BACKGROUND

Athletes are twice as likely to rupture the anterior cruciate ligament (ACL) on their healthy contralateral knee than the reconstructed graft after ACL reconstruction (ACLR). Although physical testing is commonly used after ACLR to assess injury risk to the operated knee, strength, jump, and change-of-direction performance and biomechanical measures have not been examined in those who go on to experience a contralateral ACL injury, to identify factors that may be associated with injury risk.

PURPOSE

To prospectively examine differences in biomechanical and clinical performance measures in male athletes 9 months after ACLR between those who ruptured their previously uninjured contralateral ACL and those who did not at 2-year follow-up and to examine the ability of these differences to predict contralateral ACL injury.

STUDY DESIGN

Case-control study; Level of evidence, 3.

METHODS

A cohort of male athletes returning to level 1 sports after ACLR (N = 1045) underwent isokinetic strength testing and 3-dimensional biomechanical analysis of jump and change-of-direction tests 9 months after surgery. Participants were followed up at 2 years regarding return to play or at second ACL injury. Between-group differences were analyzed in patient-reported outcomes, performance measures, and 3-dimensional biomechanics for the contralateral limb and asymmetry. Logistic regression was applied to determine the ability of identified differences to predict contralateral ACL injury.

RESULTS

Of the cohort, 993 had follow-up at 2 years (95%), with 67 experiencing a contralateral ACL injury and 38 an ipsilateral injury. Male athletes who had a contralateral ACL injury had lower quadriceps strength and biomechanical differences on the contralateral limb during double- and single-leg drop jump tests as compared with those who did not experience an injury. Differences were related primarily to deficits in sagittal plane mechanics and plyometric ability on the contralateral side. These variables could explain group membership with fair to good ability (area under the curve, 0.74-0.80). Patient-reported outcomes, limb symmetry of clinical performance measures, and biomechanical measures in change-of-direction tasks did not differentiate those at risk for contralateral injury.

CONCLUSION

This study highlights the importance of sagittal plane control during drop jump tasks and the limited utility of limb symmetry in performance and biomechanical measures when assessing future contralateral ACL injury risk in male athletes. Targeting the identified differences in quadriceps strength and plyometric ability during late-stage rehabilitation and testing may reduce ACL injury risk in healthy limbs in male athletes playing level 1 sports.

CLINICAL RELEVANCE

This study highlights the importance of assessing the contralateral limb after ACLR and identifies biomechanical differences, particularly in the sagittal plane in drop jump tasks, that may be associated with injury to this limb. These factors could be targeted during assessment and rehabilitation with additional quadriceps strengthening and plyometric exercises after ACLR to potentially reduce the high risk of injury to the previously healthy knee.

REGISTRATION

NCT02771548 (ClinicalTrials.gov identifier).

Exploring the Application of Pattern Recognition and Machine Learning for Identifying Movement Phenotypes During Deep Squat and Hurdle Step Movements

Background

Movement screens are increasingly used in sport and rehabilitation to evaluate movement competency. However, common screens are often evaluated using subjective visual detection of a priori prescribed discrete movement features (e.g., spine angle at maximum squat depth) and may not account for whole-body movement coordination, or associations between different discrete features.

Objective

To apply pattern recognition and machine learning techniques to identify whole-body movement pattern phenotypes during the performance of exemplar functional movement screening tasks; the deep squat and hurdle step. Additionally, we also aimed to compare how discrete kinematic measures, commonly used to score movement competency, differed between emergent groups identified via pattern recognition and machine learning.

Methods

Principal component analysis (PCA) was applied to 3-dimensional (3D) trajectory data from participant's deep squat (DS) and hurdle step performance, identifying emerging features that describe orthogonal modes of inter-trial variance in the data. A gaussian mixture model (GMM) was fit and used to cluster the principal component scores as an unsupervised machine learning approach to identify emergent movement phenotypes. Between group features were analyzed using a one-way ANOVA to determine if the objective classifications were significantly different from one another.

Results

Three clusters (i.e., phenotypes) emerged for the DS and right hurdle step (RHS) and 4 phenotypes emerged for the left hurdle step (LHS). Selected discrete points commonly used to score DS and hurdle step movements were different between emergent groups. In regard to the select discrete kinematic measures, 4 out of 5, 7 out of 7 and 4 out of 7, demonstrated a main effect (p < 0.05) between phenotypes for the DS, RHS, and LHS respectively.

Conclusion

Findings support that whole-body movement analysis, pattern recognition and machine learning techniques can objectively identify movement behavior phenotypes without the need to a priori prescribe movement features. However, we also highlight important considerations that can influence outcomes when using machine learning for this purpose.

Could knee joint mechanics during the golf swing be contributing to chronic knee injuries in professional golfers?

Full three-dimensional movements and external moments in golfers' knees and the possible involvement in injuries have not been evaluated using motion capture at high sample frequencies. This study measured joint angles and external moments around the three anatomical axes in both knees of 10 professional golfers performing golf drives whilst standing on two force plates in a motion capture laboratory. Significant differences were found in the knee joint moments between the lead and trail limbs for the peak values and throughout all stages during the swing phase. A significantly higher net abduction moment impulse was seen in the trail limb compared with the lead limb (-0.518 vs. -0.135 Nms.kg^-1), indicating greater loading over the whole swing, which could contribute to knee lateral compartment or anterior cruciate ligament injuries. A significant correlation (r = -0.85) between clubhead speed at ball contact and maximum joint moment was found, with the largest correlations being found for joint moments at the top of the backswing event and at the end of the follow-through. Therefore, although knee moments can contribute to high clubhead speeds, the large moments and impulses suggest that they may also contribute to chronic knee injuries or exacerbate existing conditions.

Foot strike alters ground reaction force and knee load when stepping down during ongoing walking

BACKGROUND

When stepping down from a raised surface, either a toe or heel contact strategy is performed. Increased vertical momentum is likely to be experienced during a step descent, yet the extent to which these descent strategies influence the development of load at the ground and knee has not been examined.

RESEARCH QUESTION

Does descent strategy influence ground and knee joint loading? Does the contribution from leading and trailing limb joint mechanics differ between descent strategies?

METHODS

Twenty-two healthy male participants (age: 34.0 ± 6.5 years, height: 179 ± 6.3 cm, mass: 83.5 ± 13 kg) walked along a raised platform, stepped down from a 14 cm height utilising either a toe (n = 10) or heel (n = 12) initial contact, and continued walking. Vertical ground reaction forces and knee external adduction and flexor moments were extracted for the duration of the braking phase. Joint work was calculated for the ankle, knee, and hip in both the leading and trailing limbs.

RESULTS

Waveform analysis of the loading features indicated that a toe-contact strategy resulted in significantly reduced loading rates during early braking (1-32% of the braking phase) and significantly increased magnitude in late braking (55-96% of the braking phase). Individuals performing toe landings completed 33% greater overall work (p = 0.091) in the lead limb and utilised the lead limb ankle joint as the main shock absorber (79% of total lead limb work). Concurrently, the trailing limb performed 29% and 21% less work when lowering the centre of mass and propulsion, respectively, compared to a heel landing.

SIGNIFICANCE

A toe-contact strategy results in reduced limb and knee joint loading rates through greater utilisation of the lead limb ankle joint. A heel-contact strategy, however, can reduce loading during late braking by utilising the functionality of the trailing limb.

Athletic groin pain patients and healthy athletes demonstrate consistency in their movement strategy selection when performing multiple repetitions of a change of direction test

OBJECTIVES

To report the consistency in movement strategy selection in athletic groin pain patients and to assess whether there are differences in consistency between athletic groin pain patients and healthy athletes.

DESIGN

Cross sectional exploratory study.

METHODS

Twenty athletic groin pain patients and 21 healthy athletes performed 15 repetitions of 110° change of direction task. Lower limb and trunk kinematics alongside ground reaction forces were collected. A correlation-to-mean algorithm was used to allocate each trial to a movement strategy using kinematic and kinetic features. Mann-Whitney U tests were used to compare the frequency of the most selected strategy (i.e. consistency) and fuzziness between athletic groin pain patients and healthy athletes. Chi-squared tests were used to compare the strategy selection between athletic groin pain patients and healthy athletes.

RESULTS

There were no differences between groups in consistency in movement strategy selection (>80%). Athletic groin pain patients tended to select a knee dominant movement strategy whereas healthy athletes preferred an ankle dominant movement strategy.

CONCLUSIONS

The consistency observed in athletic groin pain patients supports the implementation of movement strategy assessments to inform AGP rehabilitation programmes tailored to athletes' deficiencies. Such assessments could help enhance the success of athletic groin pain rehabilitation. Differences in movement strategy selection might not be associated with injury state since there were no differences between athletic groin pain patients and healthy athletes.

Objective classification and scoring of movement deficiencies in patients with anterior cruciate ligament reconstruction

Motion analysis systems are widely employed to identify movement deficiencies—e.g. patterns that potentially increase the risk of injury or inhibit performance. However, findings across studies are often conflicting in respect to what a movement deficiency is or the magnitude of association to a specific injury. This study tests the information content within movement data using a data driven framework that was taught to classify movement data into the classes: NORM, ACL^OP and ACL^{NO OP}, without the input of expert knowledge. The NORM class was presented by 62 subjects (124 NORM limbs), while 156 subjects with ACL reconstruction represented the ACL^OP and ACL^{NO OP} class (156 limbs each class). Movement data from jumping, hopping and change of direction exercises were examined, using a variety of machine learning techniques. A stratified shuffle split cross-validation was used to obtain a measure of expected accuracy for each step within the analysis. Classification accuracies (from best performing classifiers) ranged from 52 to 81%, using up to 5 features. The exercise with the highest classification accuracy was the double leg drop jump (DLDJ; 81%), the highest classification accuracy when considering only the NORM class was observed in the single leg hop (81%), while the DLDJ demonstrated the highest classification accuracy when considering only for the ACL^OP and ACL^{NO OP} class (84%). These classification accuracies demonstrate that biomechanical data contains valuable information and that it is possible to differentiate normal from rehabilitating movement patterns. Further, findings highlight that a few features contain most of the information, that it is important to seek to understand what a classification model has learned, that symmetry measures are important, that exercises capture different qualities and that not all subjects within a normative cohort utilise ‘true’ normative movement patterns (only 27 to 71%).