Dataset Information

Motion Sensors-Based Machine Learning Approach for the Identification of Anterior Cruciate Ligament Gait Patterns in On-the-Field Activities in Rugby Players.

ABSTRACT: Anterior cruciate ligament (ACL) injuries are common among athletes. Despite a successful return to sport (RTS) for most of the injured athletes, a significant proportion do not return to competitive levels, and thus RTS post ACL reconstruction still represents a challenge for clinicians. Wearable sensors, owing to their small size and low cost, can represent an opportunity for the management of athletes on-the-field after RTS by providing guidance to associated clinicians. In particular, this study aims to investigate the ability of a set of inertial sensors worn on the lower-limbs by rugby players involved in a change-of-direction (COD) activity to differentiate between healthy and post-ACL groups via the use of machine learning. Twelve male participants (six healthy and six post-ACL athletes who were deemed to have successfully returned to competitive rugby and tested in the 5-10 year period following the injury) were recruited for the study. Time- and frequency-domain features were extracted from the raw inertial data collected. Several machine learning models were tested, such as k-nearest neighbors, naïve Bayes, support vector machine, gradient boosting tree, multi-layer perceptron, and stacking. Feature selection was implemented in the learning model, and leave-one-subject-out cross-validation (LOSO-CV) was adopted to estimate training and test errors. Results obtained show that it is possible to correctly discriminate between healthy and post-ACL injury subjects with an accuracy of 73.07% (multi-layer perceptron) and sensitivity of 81.8% (gradient boosting). The results of this study demonstrate the feasibility of using body-worn motion sensors and machine learning approaches for the identification of post-ACL gait patterns in athletes performing sport tasks on-the-field even a number of years after the injury occurred.

SUBMITTER: Tedesco S

PROVIDER: S-EPMC7309071 | biostudies-literature | 2020 May

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Motion Sensors-Based Machine Learning Approach for the Identification of Anterior Cruciate Ligament Gait Patterns in On-the-Field Activities in Rugby Players.

Tedesco Salvatore S Crowe Colum C Ryan Andrew A Sica Marco M Scheurer Sebastian S Clifford Amanda M AM Brown Kenneth N KN O'Flynn Brendan B

Sensors (Basel, Switzerland) 20200527 11

Anterior cruciate ligament (ACL) injuries are common among athletes. Despite a successful return to sport (RTS) for most of the injured athletes, a significant proportion do not return to competitive levels, and thus RTS post ACL reconstruction still represents a challenge for clinicians. Wearable sensors, owing to their small size and low cost, can represent an opportunity for the management of athletes on-the-field after RTS by providing guidance to associated clinicians. In particular, this s ...[more]

PMID: 32471051

Similar Datasets

Project description:BackgroundAnterior cruciate ligament (ACL) injuries represent a significant burden to rugby players. Improving our understanding of the patterns and biomechanics that result in ACL injury may aid in the design of effective prevention programs.PurposeTo describe, using video analysis, the mechanisms, situational patterns, and biomechanics of ACL injuries in professional rugby matches. Further aims were to document injuries according to pitch location and timing within the match.Study designCase series; Level of evidence, 4.MethodsA total of 62 ACL injuries were identified in players of the 4 most important rugby leagues across 4 consecutive seasons. We analyzed 57 (92%) injury videos for injury mechanism and situational patterns; biomechanical analysis was performed on indirect and noncontact ACL injuries only (38 cases available). Three reviewers independently evaluated each video.ResultsMore injuries occurred while attacking than defending (41 [72%] vs 16 [28%]; P < .01). Regarding mechanism, 18 (32%) injuries were direct contact; 15 (26%), indirect contact; and 24 (42%), noncontact. Most direct contact injuries involved being tackled directly to the knee (n = 10). Three situational patterns were identified for players who had a noncontact or indirect contact injury: offensive change of direction (COD) (n = 18), being tackled (n = 10), and pressing/tackling (n = 8). Injuries generally involved a knee-loading strategy in the sagittal plane, which was accompanied by knee valgus loading in most cases (94%). Overall, 73% of injuries occurred during the first 40 minutes of effective playing time.ConclusionMost ACL injuries in professional male rugby players happened through a noncontact or indirect contact mechanism (68%). Three situational patterns were described, including offensive change of direction, being tackled, and pressing/tackling. Biomechanical analysis confirmed a multiplanar mechanism, with a knee-loading pattern in the sagittal plane accompanied by dynamic valgus. As most injuries occurred in the first 40 minutes, accumulated fatigue appears not to be a major risk factor for ACL injury.

Project description:BACKGROUND:The burden of anterior cruciate ligament (ACL) injury in professional soccer players is particularly relevant as it represents a potentially career-threatening injury. HYPOTHESIS:Our hypotheses were that (1) injury incidence rate would be similar to that reported in the literature, (2) we would identify a uniform distribution of the injuries along the season, and (3) injury incidence rate would be similar in high-ranked and lower ranked teams, based on final placement in the league. STUDY DESIGN:Descriptive epidemiological study. LEVEL OF EVIDENCE:Level 4. METHODS:Professional male soccer players participating in the Serie A championship league in 7 consecutive seasons (2011-2012 to 2017-2018) were screened to identify ACL injuries through the online football archive transfermarkt.com . Exposure in matches and training were calculated. RESULTS:There were 84 ACL injuries found (mean player age, 25.3 ± 4.2 years). Overall, 25% of ACL injuries were reruptures (15%) or contralateral injuries (10%). ACL incidence rate was 0.4215 per 1000 hours of play during Serie A matches, 0.0305 per 1000 hours of training (rate ratio [RR], 13.8; 95% CI, 8.4-22.7; P < 0.0001), and 0.0618 per 1000 hours of total play. Injury distribution had a bimodal peak, with the highest number of events in October and March. Alternatively, training injuries peaked in June and July. A significantly higher incidence rate was found for the teams ranked from 1st to 4th place compared with those ranked 5th to 20th (0.1256 vs 0.0559 per 1000 hours of play; RR, 2.2; 95% CI, 1.4-3.6; P = 0.0003). A similar finding was found for injury incidence proportion (3.76% vs 1.64%; P = 0.0003). CONCLUSION:The overall incidence rate of ACL injuries in Italian Serie A was 0.062 per 1000 hours, with a 14-fold risk in matches compared with training. Relevantly, 25% were second injuries. Most injuries occurred in October and March, and an almost 2-fold incidence rate and incidence proportion were noted in those teams ranked in the first 4 positions of the championship league. CLINICAL RELEVANCE:Knowing the precise epidemiology of ACL injury in one of the most competitive professional football championship leagues could help delineate fields of research aimed to investigate its risk factors.

Project description:ContextGait biomechanics are linked to biochemical changes that contribute to the development of posttraumatic knee osteoarthritis in individuals with anterior cruciate ligament reconstruction (ACLR). It remains unknown if modifying peak loading during gait using real-time biofeedback will result in acute biochemical changes related to cartilage metabolism.ObjectiveTo determine if acutely manipulating peak vertical ground reaction force (vGRF) during gait influences acute changes in serum cartilage oligomeric matrix protein concentration (sCOMP) among individuals with ACLR.DesignCrossover study.Patients or other participantsThirty individuals with unilateral ACLR participated (70% female, age = 20.43 ± 2.91 years old, body mass index = 24.42 ± 4.25, months post-ACLR = 47.83 ± 26.97). Additionally, we identified a subgroup of participants who demonstrated an increase in sCOMP after the control or natural loading condition (sCOMPCHANGE > 0 ng/mL, n = 22, 70% female, age = 20.32 ± 3.00 years old, body mass index = 24.73 ± 4.33, months post-ACLR = 47.27 ± 29.32).Main outcome measure(s)Serum was collected both prior to and immediately after each condition to determine sCOMPchange.Intervention(s)All participants attended 4 sessions that involved 20 minutes of walking on a force-measuring treadmill consisting of a control condition (natural loading) followed by random ordering of 3 loading conditions with real-time biofeedback: (1) symmetric vGRF between limbs, (2) a 5% increase in vGRF (high loading) and (3) a 5% decrease in vGRF (low loading). A general linear mixed model was used to determine differences in sCOMPCHANGE between altered loading conditions and the control group in the entire cohort and the subgroup.ResultsThe sCOMPCHANGE was not different across loading conditions for the entire cohort (F3,29 = 1.34, P = .282). Within the subgroup, sCOMPCHANGE was less during high loading (1.95 ± 24.22 ng/mL, t21 = -3.53, P = .005) and symmetric loading (9.93 ± 21.45 ng/mL, t21 = -2.86, P = .025) compared with the control condition (25.79 ± 21.40 ng/mL).ConclusionsIncreasing peak vGRF during gait decreased sCOMP in individuals with ACLR who naturally demonstrated an increase in sCOMP after 20 minutes of walking.Trial registryClinicalTrials.gov (NCT03035994).

Project description:PurposeTo compare gait biomechanics throughout stance phase 6 and 12 months after unilateral anterior cruciate ligament reconstruction (ACLR) between ACLR and contralateral limbs and compared with controls.MethodsVertical ground reaction force (vGRF), knee flexion angle (KFA), and internal knee extension moment (KEM) were collected bilaterally 6 and 12 months post-ACLR in 30 individuals (50% female, 22 ± 3 yr, body mass index = 23.8 ± 2.2 kg·m) and at a single time point in 30 matched uninjured controls (50% female, 22 ± 4 yr, body mass index = 23.6 ± 2.1 kg·m). Functional analyses of variance were used to evaluate the effects of limb (ACLR, contralateral, and control) and time (6 and 12 months) on biomechanical outcomes throughout stance.ResultsCompared with the uninjured controls, the ACLR group demonstrated bilaterally lesser vGRF (ACLR, 9% body weight [BW]; contralateral, 4%BW) during early stance and greater vGRF during midstance (ACLR, 5%BW; contralateral, 4%BW) 6 months post-ACLR. Compared to the uninjured controls, the ACLR group demonstrated bilaterally lesser vGRF (ACLR, 10%BW; contralateral, 8%BW) during early stance and greater vGRF during midstance (ACLR, 5%BW; contralateral, 5%BW) 12 months post-ACLR. Compared with controls, the ACLR limb demonstrated lesser KFA during early stance at 6 (2.3°) and 12 months post-ACLR (2.0°), and the contralateral limb demonstrated lesser KFA during early stance at 12 months post-ACLR (2.8°). Compared with controls, the ACLR limb demonstrated lesser KEM during early stance at both 6 months (0.011BW × height) and 12 months (0.007BW × height) post-ACLR, and the contralateral limb demonstrated lesser KEM during early stance only at 12 months (0.006BW × height).ConclusionsWalking biomechanics are altered bilaterally after ACLR. During the first 12 months post-ACLR, both the ACLR and contralateral limbs demonstrate biomechanical differences compared with control limbs. Differences between the contralateral and control limbs increase from 6 to 12 months post-ACLR.

Dataset Information

Motion Sensors-Based Machine Learning Approach for the Identification of Anterior Cruciate Ligament Gait Patterns in On-the-Field Activities in Rugby Players.

Publications

Motion Sensors-Based Machine Learning Approach for the Identification of Anterior Cruciate Ligament Gait Patterns in On-the-Field Activities in Rugby Players.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets