Injury prediction in veteran football players using the Functional Movement Screen™

A Narrative Review for a Machine Learning Application in Sports: An Example Based on Injury Forecasting in Soccer

In the last decade, the number of studies about machine learning algorithms applied to sports, e.g., injury forecasting and athlete performance prediction, have rapidly increased. Due to the number of works and experiments already present in the state-of-the-art regarding machine-learning techniques in sport science, the aim of this narrative review is to provide a guideline describing a correct approach for training, validating, and testing machine learning models to predict events in sports science. The main contribution of this narrative review is to highlight any possible strengths and limitations during all the stages of model development, i.e., training, validation, testing, and interpretation, in order to limit possible errors that could induce misleading results. In particular, this paper shows an example about injury forecaster that provides a description of all the features that could be used to predict injuries, all the possible pre-processing approaches for time series analysis, how to correctly split the dataset to train and test the predictive models, and the importance to explain the decision-making approach of the white and black box models.

Systematic review of screening tools for common soccer injuries and their risk factors

Background

Several screening tools are available for use in a clinical setting to predict injury. However, there is a lack of evidence regarding the accuracy of these tools to predict soccer-specific injuries.

Objectives

The purpose of this systematic literature review was to determine the psychometric properties or accuracy of screening tools for common soccer injuries.

Methods

A systematic review of diagnostic test accuracy was undertaken based on the Joanna Briggs Institute (JBI) procedure for conducting systematic reviews. Databases such as SPORT Discus, Cinahl, Medline, Science Direct, PubMed and grey literature were searched in order to access suitable studies.

Results

A total of 10 studies were included for the analysis – three were analysed quantitatively whilst the remaining seven were analysed qualitatively. The screening tools were of high reliability, sensitivity and specificity (calculated as intraclass correlation coefficient [ICC] (0.68 95% confidence interval [CI]: 0.52–0.84 and 0.64 95% CI: 0.61–0.66, respectively).

Conclusion

The screening tools assessed for the prediction of common soccer injuries that emerged from this systematic review include the Functional Movement Screening (FMS™), the Landing Error Scoring System (LESS), the Tuck Jump Assessment, the Soccer Injury Movement Screening (SIMS) and the conventional hamstrings to quadriceps ratio; all with good evidence of predicting common soccer injuries. These tools were of high sensitivity and specificity thus reliable for soccer screening.

Clinical implications

The validity of these tools is acceptable and therefore the authors recommend that these tools be included in an injury prevention programme for soccer players.

Factors Influencing the Relationship Between the Functional Movement Screen and Injury Risk in Sporting Populations: A Systematic Review and Meta-analysis

BACKGROUND

Studies investigating the association between the Functional Movement Screen (FMS) and sports injury risk have reported mixed results across a range of athlete populations.

OBJECTIVES

The purpose of this systematic review was to identify whether athlete age, sex, sport type, injury definition and mechanism contribute to the variable findings.

STUDY DESIGN

Systematic review and meta-analysis.

METHODS

A systematic search was conducted in October 2018 using PubMed, EBSCOhost, Scopus, EmBase and Web of Science databases. Studies were included if they were peer reviewed and published in English language, included athletes from any competition level, performed the FMS at baseline to determine risk groups based on FMS composite score, asymmetry or pain, and prospectively observed injury incidence during training and competition. Study eligibility assessment and data extraction was performed by two reviewers. Random effects meta-analyses were used to determine odds ratio (OR), sensitivity and specificity with 95% confidence intervals. Sub-group analyses were based on athlete age, sex, sport type, injury definition, and injury mechanism.

RESULTS

Twenty-nine studies were included in the FMS composite score meta-analysis. There was a smaller effect for junior (OR = 1.03 [0.67-1.59]; p = 0.881) compared to senior athletes (OR = 1.80 [1.17-2.78]; p = 0.008) and for male (OR = 1.79 [1.08-2.96]; p = 0.024) compared to female (OR = 1.92 [0.43-8.56]; p = 0.392) athletes. FMS composite scores were most likely to be associated with increased injury risk in rugby (OR = 5.92 [1.67-20.92]; p = 0.006), and to a lesser extent American football (OR = 4.41 [0.94-20.61]; p = 0.059) and ice hockey (OR = 3.70 [0.89-15.42]; p = 0.072), compared to other sports. Specificity values were higher than sensitivity values for FMS composite score. Eleven studies were included in the FMS asymmetry meta-analysis with insufficient study numbers to generate sport type subgroups. There was a larger effect for senior (OR = 1.78 [1.16-2.73]; p = 0.008) compared to junior athletes (OR = 1.21 [0.75-1.96]; p = 0.432). Sensitivity values were higher than specificity values for FMS asymmetry. For all FMS outcomes, there were minimal differences across injury definitions and mechanisms. Only four studies provided information about FMS pain and injury risk. There was a smaller effect for senior athletes (OR = 1.28 [0.33-4.96]; p = 0.723) compared to junior athletes (OR = 1.71 [1.16-2.50]; p = 0.006). Specificity values were higher than sensitivity values for FMS pain.

CONCLUSION

Athlete age, sex and sport type explained some of the variable findings of FMS prospective injury-risk studies. FMS composite scores and asymmetry were more useful for estimating injury risk in senior compared to junior athletes. Effect sizes tended to be small except for FMS composite scores in rugby, ice hockey and American football athletes.

PROTOCOL REGISTRATION

CRD42018092916.

Effects of dynamic neuromuscular stabilization (DNS) training on functional movements

Functional movements (FMs) dysfunction is a potential risk factor of injuries. A variety of training strategies is proposed to improve the performance of FMs. We investigated if a system of fundamental movement exercises called Dynamic Neuromuscular Stabilization (DNS) could improve FMs. Thirty-four female students were randomly assigned into two matched groups to receive DNS (the study protocol) versus physical fitness (PF) training. The groups practiced for six-weeks (three sessions of 50 min weekly). We used five FMs tests as pre and post measures of exercise effectiveness. Repeated Measures ANOVA showed a significant interaction in all five FMs tests in favour of DNS group (F_(1,32) ≥ 4.13, P ≤ .001 and ƞ² ≥ 0.29), meaning that DNS group had a higher progress rate compared to that of PF group. Based on Eta-square coefficients, the highest and lowest differences in the progression rate were observed in Y-Balance and Functional Movement Screening Tests, respectively. Our findings supported the hypothesis that fundamental movements of DNS could be used to improve FMs. However, the progression coefficient declined as FMs became more specific. Lower progression of "specific FMs" suggests that it might prove more effective to add "specific training" to "fundamental training" for them.

Functional assessment of women practising combat sports and team sports using the Functional Movement Screen

Study aim: The objective of the study was to reveal the functional limits of the motor system in women practising combat sports and team sports.

Material and methods: 102 women (mean age 25.2 years, body mass 62.2 kg, body height 168.3 cm) practising competitive ITF (International Taekwon-Do Federation) taekwon-do (n = 22), Brazilian jiu-jitsu (BJJ) (n = 15), football (n = 35) and basketball (n = 30) participated in the study. The assessment tool was the FMS test, comprising 7 movement patterns scored on a scale of 0–3.

Results: The studied female athletes earned medium scores. Women practising combat sports scored generally higher in the FMS test, although the difference was not significant (combat sports – mean value 15.57 ± 2.39, team sports – mean value 14.72 ± 1.93, difference – p = 0.07). Statistically significant differences (p < 0.05) were observed in the second (hurdle step) and the fifth pattern (active straight leg raise – ASLR). The aggregated FMS results of female taekwon-do (15.77) and BJJ athletes (15.22) were similar. Significant differences (p < 0.05) were observed in one pattern (ASLR). Women practising football (14.77) and basketball (14.67) attained a similar level of results in the test. Statistically significant differences were noted in two trials: footballers scored higher in the ASLR task (p < 0.05), and basketball players scored higher in rotary stability (p < 0.01).

Conclusion: A statistically significant result in the FMS test was obtained by martial arts athletes, which may point to a higher level of functional movement, which may be a result of more universal training.

Association of the Functional Movement Screen™ with match-injury burden in men's community rugby union

Evidence supporting use of the Functional Movement Screen (FMS^TM) to identify athletes' risk of injury is equivocal. Furthermore, few studies account for exposure to risk during analysis. This study investigated the association of FMS^TM performance with incidence and burden of match-injuries in adult community rugby players. 277 players performed the FMS^TM during pre-season and in-season time-loss injuries and match exposure were recorded. The associations between FMS^TM score, pain, and movement-pattern asymmetries with match-injury incidence (≥8-days time-loss/1000hours), severe match-injury incidence (>28-days time-loss/1000hours), and match-injury burden (total time-loss days/1000hours for ≥8-days match-injuries) were analysed using Poisson regression. Multivariate analysis indicated players with pain and movement-pattern asymmetry during pre-season had 2.9 times higher severe match-injury incidence (RR, 90%CI = 2.9, 0.9-9.7) and match-injury burden (RR, 90%CI = 2.9, 1.3-6.6). Players with a typically low FMS^TM score (mean - 1SD threshold) were estimated to have a 50% greater match-injury burden compared to players with a typically high FMS^TM score (mean + 1SD threshold) as match-injury burden was 10% lower per 1-unit increase in FMS^TM score. As the strongest association with injury outcome was found for players with pain and asymmetry, when implementing the FMS^TM it is advisable to prioritise these players for further assessment and subsequent treatment.

Utility of FMS to understand injury incidence in sports: current perspectives

The Functional Movement Screen (FMS) is a popular movement screen used by rehabilitation, as well as strength and conditioning, professionals. The FMS, like other movement screens, identifies movement dysfunction in those at risk of, but not currently experiencing, signs or symptoms of a musculoskeletal injury. Seven movement patterns comprise the FMS, which was designed to screen fundamental movement requiring a balance between stability and mobility. The 7 movement patterns are summed to a composite FMS score. For an instrument to have wide applicability and acceptability, there must be high levels of reliability, validity, and accuracy. The FMS is certainly a reliable tool, and can be consistently scored within and between raters. Although the FMS has high face and content validity, the criterion validity (discriminant and convergent) is low. Additionally, the FMS does not appear to be studying a single construct, challenging the use of the summed composite FMS score. The accuracy of the FMS in screening for injury is also suspect, with low sensitivity in almost all studies, although specificity is higher. Finally, within the FMS literature, the concepts of prediction and association are conflated, combined with flawed cohort studies, leading to questions about the efficacy of the FMS to screen for injury. Future research on the use of the FMS, either the composite score or the individual movement patterns, to screen for injury or injury risk in adequately powered, well-designed studies are required to determine if the FMS is appropriate for use as a movement screen.

Dynamic Movement Assessment and Functional Movement Screening for injury prediction: a systematic review

ABSTRACT Dynamic Movement AssessmentTM (DMATM) and Functional Movement ScreeningTM (FMSTM) are tools to predict the risk of musculoskeletal injuries in individuals who practice physical activities. This systematic review aimed to evaluate the association of DMATM and FMSTM with the risk of musculoskeletal injuries, in different physical activities, categorizing by analysis. A research without language or time filters was carried out in November 2016 in MEDLINE, Google Scholar, SciELO, SCOPUS, SPORTDiscus, CINAHL and BVS databases using the keywords: “injury prediction”, “injury risk”, “sensitivity”, “specificity”, “functional movement screening”, and “dynamic movement assessment”. Prospective studies that analyzed the association between DMATM and FMSTM with the risk of musculoskeletal injuries in physical activities were included. The data extracted from the studies were: participant’s profile, sample size, injury’s classification criteria, follow-up time, and the results presented, subdivided by the type of statistical analysis. The risk of bias was performed with Newcastle-Ottawa Scale for cohort studies. No study with DMATM was found. A total of 20 FMSTM studies analyzing one or more of the following indicators were included: diagnostic accuracy (PPV, NPV and AUC), odds ratios (OR) or relative risk (RR). FMSTM showed a sensitivity=12 to 99%; specificity=38 to 97%; PPV=25 to 91%; NPV=28 to 85%; AUC=0.42 to 0.68; OR=0.53 to 54.5; and RR=0.16-5.44. The FMSTM has proven to be a predictor of musculoskeletal injuries. However, due to methodological limitations, its indiscriminate usage should be avoided.

Soccer Injury Movement Screen (SIMS) Composite Score Is Not Associated With Injury Among Semiprofessional Soccer Players

Background The association between movement quality and injury is equivocal. No soccer-specific movement assessment has been prospectively investigated in relation to injury risk. Objectives To investigate the association between a soccer-specific movement-quality assessment and injury risk among semiprofessional soccer players. Methods In this prospective cohort study, semiprofessional soccer players (n = 306) from 12 clubs completed the Soccer Injury Movement Screen (SIMS) during the preseason period. Individual training/match exposure and noncontact time-loss injuries were recorded prospectively for the entirety of the 2016 season. Relative risks were calculated, and presented with 90% confidence intervals, for the SIMS composite and individual subtest scores from generalized linear models with Poisson distribution offset for exposure. Results When considering noncontact time-loss lower extremity injuries (primary level of analysis), there was a most likely trivial association with the SIMS composite score. Similarly, the SIMS composite score demonstrated most likely to likely trivial associations with all injury categories included in the secondary level of analysis (noncontact time-loss hip/groin, thigh, knee, and ankle injuries). When considering hamstring strains and ankle sprains specifically (tertiary level of analysis), the SIMS composite score demonstrated very likely trivial associations. A total of 262 noncontact time-loss injuries were recorded. The overall (training and match exposure combined) incidence of noncontact time-loss injury was 12/1000 hours. Conclusion The SIMS composite score demonstrated no association with any of the investigated categories of soccer-related injury. The SIMS composite score should not be used to group players into high- or low-risk groups. Level of Evidence Prognosis, level 4. J Orthop Sports Phys Ther 2018;48(8):630-636. Epub 8 May 2018. doi:10.2519/jospt.2018.8037.

Changes in Muscle Thickness Across Positions on Ultrasound Imaging in Participants With or Without a History of Low Back Pain

CONTEXT

  Injury-prediction models have identified trunk muscle function as an identifiable factor for future injury. A history of low back pain (HxLBP) may also place athletes at increased risk for future low back pain. Reduced muscle thickness of the lumbar multifidus (LM) and transversus abdominis (TrA) has been reported among populations with clinical low back pain via ultrasound imaging in multiple positions. However, the roles of the LM and TrA in a more functional cohort and for injury prediction are still unknown.

OBJECTIVES

  To (1) assess the reliability of LM and TrA ultrasound measures, (2) compare changes in muscle thickness across positions between persons reporting or not reporting HxLBP, and (3) determine the ability to distinguish between groups.

DESIGN

  Cross-sectional study.

SETTING

  Research laboratory.

PATIENTS OR OTHER PARTICIPANTS

  Participants were 34 people who did not report HxLBP (age = 22 ± 7 years, body mass index = 23.7 ± 2.7) and 25 people who reported HxLBP (age = 25 ± 10 years, body mass index = 24.0 ± 3.2).

MAIN OUTCOME MEASURE(S)

  Muscle thickness and changes in muscle thickness of the LM and TrA as shown on ultrasound imaging.

RESULTS

  Intraclass correlation coefficients ranged from 0.641 to 0.943 for all thickness measures and from 0 to 0.693 for all averaged thickness modulations bilaterally. Participants who reported HxLBP had voluntarily reduced TrA thickness modulations compared with those not reporting HxLBP ( P = .03), and the testing position influenced TrA thickness modulations ( P < .01). No differences were observed for LM thickness modulations between groups or positions ( P > .05). A tabletop cutoff value of 1.32 had a sensitivity of 0.640 and a specificity of 0.706, whereas a seated cutoff value of 1.18 had a sensitivity of 0.600 and a specificity of 0.647.

CONCLUSIONS

  In participants reporting HxLBP, TrA thickness modulations were lower and both tabletop and seated thickness modulations were able to distinguish reported HxLBP status. These findings suggest that TrA muscle function may be altered by HxLBP.

Multicomponent Musculoskeletal Movement Assessment Tools: A Systematic Review and Critical Appraisal of Their Development and Applicability to Professional Practice

Multicomponent movement assessment tools have become commonplace to measure movement quality, proposing to indicate injury risk and performance capabilities. Despite popular use, there has been no attempt to compare the components of each tool reported in the literature, the processes in which they were developed, or the underpinning rationale for their included content. As such, the objective of this systematic review was to provide a comprehensive summary of current movement assessment tools and appraise the evidence supporting their development. A systematic literature search was performed using PRISMA guidelines to identify multicomponent movement assessment tools. Commonalities between tools and the evidence provided to support the content of each tool was identified. Each tool underwent critical appraisal to identify the rigor in which it was developed, and its applicability to professional practice. Eleven tools were identified, of which 5 provided evidence to support their content as assessments of movement quality. One assessment tool (Soccer Injury Movement Screen [SIMS]) received an overall score of above 65% on critical appraisal, with a further 2 tools (Movement Competency Screen [MCS] and modified 4 movement screen [M4-MS]) scoring above 60%. Only the MCS provided clear justification for its developmental process. The remaining 8 tools scored between 40 and 60%. On appraisal, the MCS, M4-MS, and SIMS seem to provide the most practical value for assessing movement quality as they provide the strongest reports of developmental rigor and an identifiable evidence base. In addition, considering the evidence provided, these tools may have the strongest potential for identifying performance capabilities and guiding exercise prescription in athletic and sport-specific populations.