BackgroundHead-on-head impacts are a risk factor for concussion, which is a concern for sports. Computer vision frameworks may provide an automated process to identify head-on-head impacts, although this has not been applied or evaluated in rugby.MethodsThis study developed and evaluated a novel computer vision framework to automatically classify head-on-head and non-head-on-head impacts. Tackle events from professional rugby league matches were coded as either head-on-head or non-head-on-head impacts. These included non-televised standard-definition and televised high-definition video clips to train (n=341) and test (n=670) the framework. A computer vision framework consisting of two deep learning networks, an object detection algorithm and three-dimensional Convolutional Neural Networks, was employed and compared with the analyst-coded criterion. Sensitivity, specificity and positive predictive value were reported.ResultsThe overall performance evaluation of the framework to classify head-on-head impacts against manual coding had a sensitivity, specificity and positive predictive value (95% CIs) of 68% (58% to 78%), 84% (78% to 88%) and 0.61 (0.54 to 0.69) in standard-definition clips, and 65% (55% to 75%), 84% (79% to 89%) and 0.61 (0.53 to 0.68) in high-definition clips.ConclusionThe study introduces a novel computer vision framework for head-on-head impact detection. Governing bodies may also use the framework in real time, or for retrospective analysis of historical videos, to establish head-on-head rates and evaluate prevention strategies. Future work should explore the application of the framework to other head-contact mechanisms and also the utility in real time to identify potential events for clinical assessment.
OBJECTIVES:To describe and compare biceps femoris long head (BFlh) muscle architecture between limbs with and without a recent history of T-junction hamstring injury. DESIGN:Case-control study. SETTING:English Premier League club. MAIN OUTCOME MEASURES:Participants were 30 professional male soccer players, including 5 cases with recent history of unilateral T-junction hamstring injury and 25 controls with no recent history of any hamstring injury. Participants had their BFlh fascicle length (FL), pennation angle (PA) and muscle thickness (MT) assessed bilaterally using wide-field-of-view ultrasound. Paired t-tests compared FL, PA and MT between previously injured (left) BFlh and contralateral uninjured (right) BFlh within cases. Un-paired t-tests compared left BFlh - right BFlh (asymmetry) in these measures between the case and control groups. RESULTS:Within cases, MT was significantly less in previously injured compared to contralateral uninjured BFlh (p < 0.01; mean paired difference [95%CI] = -0.48 cm [-0.59 cm to -0.36 cm]). Asymmetry in BFlh muscle thickness was significantly larger in the case group compared to control group (p < 0.01; between-group mean difference [95%CI] = -0.51 cm [-0.64 cm to -0.37 cm]). BFlh FL and PA did not differ significantly within cases, or between the case and control groups. CONCLUSION:Between-limb deficits in BFlh MT exist following T-junction hamstring injury.
This study analysed skill efficacy in relation to successful rally, set and match outcomes in elite men's volleyball, with consideration given to the potential impact of complexes. Footage from 26 pool-stage matches, across 232 players, of the 2022 F & eacute;d & eacute;ration Internationale de Volleyball (FIVB) Men's World Championships were coded. Mixed-effects logistic regression models assessed how one-unit increases in skill efficacy influenced the odds of success at each outcome level. All skills positively influenced rally, set and match outcomes, except for block efficacy, which demonstrated uncertainty at the match level. Considering odds ratios, spike efficacy exerted the greatest effect on rally outcome (5.73), whilst setting was most influential (1.43) and for match outcome (1.35) along with serve receive (1.34). The conversion of odds ratios to predicted probability and subsequent analysis provided novel insight into the magnitude of skill efficacy increments on success likelihood. Regarding complexes, only the slope of efficacy for defence varied significantly across complexes for both rally (p < 0.01) and match (p = 0.04) outcomes. These findings highlight the importance of aligning skill development priorities with the specific performance outcome level, supporting efficacy-based, context-sensitive performance analysis frameworks for elite volleyball coaching, with implications for talent identification and development.
ObjectiveQuantify between-match, -player and -team variability and compare whole- and peak-match locomotor characteristics between positions in elite female academy soccer.MethodFoot-mounted inertial measurement unit data were collected from 11 Women's Super League Academy teams (n = 257 players; 171 matches). Differences between positions and variability were analysed using partial least squares correlation analysis (PLSCA) and linear mixed effects. Two latent variables were computed as composite scores of either whole match or peak intensity variables from the PLSCA.ResultsBetween-match variability of whole- and peak-match locomotor characteristics were similar (2 to 24% vs 0.2 to 22%). Between-team and -player variability was higher for whole- than peak-match locomotor characteristics (1 to 20% vs 0.1 to 3%, 8 to 112% vs 0.3 to 77%). From 30 pairwise comparisons, there were two large effect size (ES) differences (p < 0.001), WM had greater whole and peak match composite intensity than CDM. There were 10 moderate ES differences (p < 0.05), with WM greater than CD and CAM, F greater than CD and CDM and WD greater than CDM. All other comparisons were non-significant, trivial or small.ConclusionWhole- and peak- match locomotor characteristics are similar across all outfield positions in elite female academy soccer. Between-match variability was greater for sprint distance than all other locomotor characteristics. Low variability between teams for peak locomotor characteristics means practitioners can be confident using peak reference values from this study and use them to evaluate training drill intensities of tactical-technical drills.
The aim was to develop and validate an individualised internal training exposure method by deriving weighting factors for each heart rate (HR) from detrended fluctuation analysis of heart rate variability (DFA-α1) during a graded exercise test. Thirty-seven participants (17 females; 32.72 ± 9.26 years; maximal oxygen uptake, V̇ O2max = 48.32 ± 7.95 mL kg−1 min−1) completed a step- and a ramp incremental test to measure blood lactate (BLa), DFA-α1, and cardiorespiratory fitness (CRF) variables, i.e. speed at lactate, ventilatory thresholds (LTs/VTs), and V̇ O2max. Exponential fitting of the fractional elevation of HR (ΔHR) with BLa (individualised training impulse; iTRIMP) or DFA-α1 (αTRIMP) generated individualised coefficients for both methods. The TRIMP weightings were interpolated values of BLa or DFA-α1 derived at each ΔHR through coefficients to represent individualised physiological intensity. Principal component regression evaluated the relationship between combined CRF variables and the TRIMP coefficients or weightings. Large inter-individual variation was observed at the same physiological thresholds (ΔHR at LT1/VT1 = 0.51–0.83 and LT2/VT2 = 0.63–0.96), underscoring the need for TRIMP methods to weight ΔHR and account for different exposure at similar intensity. CRF had a moderate relationship with coefficients for iTRIMP and αTRIMP methods (R2average = 0.52–0.67), but a moderate to strong relationship with their weightings at a fixed ΔHR (R2average = 0.67–0.78). αTRIMP is a valid and practically accessible method for quantifying internal training exposure using ECG-based HR monitors, which individualises physiological intensity through DFA-α1-derived weightings among individuals of varied fitness exercising at same percentages of HR.
Purpose : Investigate the concurrent agreement and test–retest reliability of 10-Hz global-positioning-system (GPS) device against a criterion measure (47-Hz radar device) to assess maximal horizontal deceleration ability (maximum deceleration [DEC Max ], average deceleration [DEC Ave ], time to stop, and distance to stop). Methods : Thirty-two male elite youth academy soccer players (age 18.1 [1.6] y, body mass 76.6 [7.9] kg) completed the acceleration–deceleration ability test with 16 completing a second test to assess test–retest reliability. Maximal horizontal deceleration ability was measured concurrently using GPS Raw (10-Hz data), GPS Export (STATSports software), and a radar device. Bland–Altman method and equivalence testing assessed concurrent agreement and intraclass correlations with coefficient of variation (%) was used to assess test–retest reliability. Results: Equivalence testing showed mean difference between the radar device and GPS-derived values of DEC Ave and DEC Max were within equivalence bounds. GPS Raw and GPS Export derived values of DEC Max showed good overall (intraclass correlations = .84–.86, coefficient of variation % = 4.50–5.48) test–retest reliability. Conclusion : Practitioners can consider using deceleration variables (DEC Ave and DEC Max ) obtained from GPS as a cost-effective, valid, and reliable alternative to radar technology to assess maximal horizontal deceleration ability in team-sport players.
This study aimed to describe the incidence of head acceleration events (HAEs) during pitch-based in-season training and matches in professional male rugby league. Data were recorded using instrumented mouthguards from 108 players (70 forwards and 38 backs) at nine Super League teams (2024 season), resulting in 468 player-training sessions and 665 player-matches included. Peak linear and angular acceleration were calculated from each HAE and analyzed using generalized linear mixed-effects models. During the 468 player-training sessions, 814 HAEs above the lowest magnitude threshold (5 g and 400 rad.s-2) were observed and the mean HAE incidence rate per player-hour was 1.52 (95% confidence intervals; 1.34-1.70). This was substantially lower than matches (25.78 [23.28-28.27] per player-hour) with HAE incidence being 17 times greater during matches compared to training (incidence rate ratio 16.96 [14.92-19.01]). Higher magnitude HAEs had a lower incidence in both training and matches (e.g., > 25 g 0.04 [0.02-0.06] and 2.01 [1.79-2.24] per player-hour). Out of 468 player-training sessions, 307 (~66%) had no HAEs > 10 g and 441 (~94%) had no HAEs > 25 g. Overall, the incidence rates of HAEs during training were low and substantially lower than match-play. However, a small proportion of relatively high in magnitude HAEs do occur during training, which could be the target of prevention interventions in training. However, given the different HAE rates between training and matches, interventions targeting matches (e.g., law modifications or reduced exposure) would have a larger effect on reducing HAEs for players than training interventions.
Purpose: Despite the known health and wellbeing benefits of taking part in sport for children and adolescents, it is reported that sports participation declines during adolescence. The purpose of this study was to explore current organized youth sport participation rates across Europe for both males and females and update current understanding. Method: Sport participation registration data was collected for 18 sports from 27 countries. In total, participation data was collected from over 5 million young people from Under 8s (U8s) to Under 18s (U18s). Differences in the participation rates between age categories were investigated using a generalized linear mixed effects model. Results: Overall, males were four times more likely to participate in organised youth sport than females' participants, with this trend apparent across all age categories and across most sports. There was a significant decrease across sports in participation rates for males during adolescence from U14-U16 and U16-U18. There was a significant decrease in participation rates for females from U14-U16 for most sports except but an increase in participation rates from U16-U18 for 12 out of 18 sports. Soccer (1262%), wrestling (391%) and boxing (209%) were the sports that had greater male sport participation rates. In contrast, dance sports (86%) and volleyball (63%) had more female participants than males. This research shows male sports participation is significantly greater than female in youth sport across Europe. Conclusion: Furthermore, findings showed that for both male and female participants, participation rates increased from U8-U14 for the majority of sports followed by reduced participation rates during adolescence. Findings of this research can be used by national governing bodies and sporting organizations to inform youth sport participation initiatives.
Determining key performance indicators and classifying players accurately between competitive levels is one of the classification challenges in sports analytics. A recent study applied Random Forest algorithm to identify important variables to classify rugby league players into academy and senior levels and achieved 82.0% and 67.5% accuracy for backs and forwards. However, the classification accuracy could be improved due to limitations in the existing method. Therefore, this study aimed to introduce and implement feature selection technique to identify key performance indicators in rugby league positional groups and assess the performances of six classification algorithms. Fifteen and fourteen of 157 performance indicators for backs and forwards were identified respectively as key performance indicators by the correlation-based feature selection method, with seven common indicators between the positional groups. Classification results show that models developed using the key performance indicators had improved performance for both positional groups than models developed using all performance indicators. 5-Nearest Neighbour produced the best classification accuracy for backs and forwards (accuracy = 85% and 77%) which is higher than the previous method's accuracies. When analysing classification questions in sport science, researchers are encouraged to evaluate multiple classification algorithms and a feature selection method should be considered for identifying key variables.
Background: Athlete exposure to contact could be a risk factor for injury. Governing bodies should provide guidelines preventing overexposure to contact. Objectives: Describe the current contact load practices and perceptions of contact load requirements within men’s and women’s rugby league to allow the Rugby Football League (RFL) to develop contact load guidelines. Methods: Participants (n=450 players, n=46 coaching staff, n=32 performance staff, n=23 medical staff) completed an online survey of 27 items, assessing the current contact load practices and perceptions within four categories: “current contact load practices” (n=12 items), “perceptions of required contact load” (n = 6 items), “monitoring of contact load” (n=3 items), and “the relationship between contact load and recovery” (n=6 items). Results: During men’s Super League pre-season, full contact and controlled contact training was typically undertaken for 15-30 minutes per week, and wrestling training for 15-45 minutes per week. During the in-season, these three training types were all typically undertaken for 15-30 mins per week. In women’s Super League, all training modalities were undertaken for up to 30 minutes per week in the pre- and in-season periods. Both men’s and women’s Super League players and staff perceived 15-30 minutes of full contact training per week was enough to prepare players for the physical demands of rugby league, but a higher duration may be required to prepare for the technical contact demands. Conclusion: Men’s and women’s Super League clubs currently undertake more contact training during pre-season than in-season, which was planned by coaches and is deemed adequate to prepare players for the demands of rugby league. This study provides data to develop contact load guidelines to improve player welfare whilst not impacting performance.
The aims were to determine the relationship(s) between match-play external load and post-match neuromuscular fatigue as latent constructs, the contribution of the specific measured variables to these latent constructs, and how these differ between forwards and backs in elite rugby union. Forty-one elite male rugby union players (22 forwards and 19 backs) from the same international rugby union team were tested, with data included from the 2020 and 2021 international seasons (11 matches; 146 player appearances). Player's match-play external loads were quantified using microtechnology (for locomotor activities) and video analysis (for collision actions). Neuromuscular fatigue was quantified using countermovement jump tests on force plates which were conducted similar to 24 to 48 hours pre- and post-match. Partial least squares correlation (PLSC) leave one variable out (LOVO) procedure established the relative variable contribution to both external load (X matrix) and neuromuscular fatigue (Y matrix) constructs. Linear mixed-effects models were then constructed to determine the variance explained by the latent scores applied to the variables representing these constructs. For external load, both locomotor and collision variables were identified for the forwards and the backs, although the identified variables differed between groups. For neuromuscular fatigue, jump height was identified as a high contributor for the forwards and the backs, with concentric impulse and reactive strength index high contributors only for the backs. The explained variance between the external load and neuromuscular fatigue latent constructs at the individual player level was 4.4% and 32.2% in the forwards and the backs models, respectively. This discrepancy may be explained by differences in match-play external loads and/or the specificity of the tests to measure indicators of fatigue. These may differ due to, for example, the activities undertaken in the different positional groups.
Rule changes within football-code team sports aim to improve performance, enhance player welfare, increase competitiveness, and provide player development opportunities. This manuscript aimed to review research investigating the effects of rule changes in football-code team sports. A systematic search of electronic databases (PubMed, ScienceDirect, CINAHL, MEDLINE, and SPORTDiscus) was performed to August 2023; keywords related to rule changes, football-code team sports, and activity type. Studies were excluded if they failed to investigate a football-code team sport, did not quantify the change of rule, or were review articles. Forty-six studies met the eligibility criteria. Four different football codes were reported: Australian rules football (n = 4), rugby league (n = 6), rugby union (n = 16), soccer (n = 20). The most common category was physical performance and match-play characteristics (n = 22). Evidence appears at a high risk of bias partly due to the quasi-experimental nature of included studies, which are inherently non-randomised, but also due to the lack of control for confounding factors within most studies included. Rule changes can result in unintended consequences to performance (e.g., longer breaks in play) and effect player behaviour (i.e., reduce tackler height in rugby) but might not achieve desired outcome (i.e., unchanged concussion incidence). Coaches and governing bodies should regularly and systematically investigate the effects of rule changes to understand their influence on performance and injury risk. It is imperative that future studies analysing rule changes within football codes account for confounding factors by implementing suitable study designs and statistical analysis techniques.
Topics in Exercise Science and Kinesiology Volume 5: Issue 1, Article 3, 2024. The rugby codes (i.e., rugby union, rugby league, rugby sevens [termed ‘rugby’]) are team-sports that impose complex physical demands upon players which in-turn, leads to domain-specific fatigue (e.g., neuromuscular, cardio-autonomic). Quantifying post-match fatigue through various methods and metrics is important to monitor player fatigue status, which influences training readiness. The specific and general barriers limiting the use of post-match fatigue monitoring in rugby are not presently known. Therefore, the aims of this study were to identify specific and general barriers (clusters of specific barriers) to the use of post-match fatigue monitoring methods and metrics in rugby across the domains of neuromuscular performance, cardio-autonomic, tissue biomarker, and self-reported fatigue, and which of these specific barriers were considered important to overcome and feasible to overcome. An international cohort of subject matter experts (SME) in rugby completed a two-round online questionnaire survey (Round One; n = 42, Round Two; n = 13), with the responses collected and analysed using Concept Mapping. Specific barrier statements were generated based on the SME responses to Round One, which were structured and then rated by the SME for importance to overcome and feasibility to overcome in Round Two. Five clusters of specific barriers (representing the general barriers) were determined based on analyses of the SME responses: 1. ‘Budget and Equipment’, 2. ‘Data and Testing Considerations’, 3. ‘Player and Coach Perceptions’, 4. ‘Test Appropriateness’, and 5. ‘Time and Space’. For both importance to overcome and feasibility to overcome, the ‘Data and Testing Considerations’ had the highest overall rating and contained the largest number of specific barriers which rated highly. These findings should be considered when identifying which post-match fatigue monitoring methods and metrics to implement in rugby, and potentially other sports.
Player movement in rugby league is complex, being spatiotemporal and multifaceted. Modeling this complexity to provide robust measures of player activity and load has proved difficult, with important aspects of player movement yet to be considered. These include the influence of time-varying covariates on player activity and the combination of different dimensions of player movement. Few studies have simultaneously categorized player activity into different activity states and investigated factors influencing the transition between states, or compared player activity and load profiles between matches and training. This study applied hidden Markov models (HMMs)—a data-driven, multivariate approach—to rugby league training and match GPS data to i) demonstrate how HMMs can combine multiple variables in a data-driven way to effectively categorize player movement states, ii) investigate the influence of two time-varying covariates, score difference and elapsed match time on player activity states, and iii) compare player activity and load profiles within and between training and match modalities. HMMs were fitted to player GPS, accelerometer and heart rate data of one English Super League team across 60 training sessions and 35 matches. Distinct activity states were detected for both matches and training, with transitions between states in matches influenced by score difference and elapsed time and clear differences in activity and load profiles between training and matches. HMMs can model the complexity of player movement to effectively profile player activity and load in rugby league and have the potential to facilitate new research across several sports.
Using a pre-post-test design, this study evaluated the impact of implementing a standard on the reliability of player management decision-making within a professional rugby union environment. Five practitioners from a High-Performance Unit (HPU) rated 22 instances of Global Positioning System (GPS)–based external training load information of 14 players across the 2021–2022 season. This rating was whether a peak/trough/normal exposure in load had occurred. The ratings were repeated at four time points (separated by 2 weeks) before (Pre 1 , Pre 2 ) and after (Post 1 , Post 2 ) implementing a consensus statement as a subjective standard (using a dashboard) developed previously within the same environment to identify peaks/troughs in player external training loads. Inter-rater agreement between individuals at each voting round was assessed using Light's Kappa, while pre-post-standard intra-rater agreement was determined from Cohen's Kappa (both with 95% confidence intervals). Changes to dashboard usability from implementing the standard were assessed by administering the System Usability Scale to 11 HPU staff at the four time points. Pre-standard moderate inter-rater agreement (Pre 1 : 0.53 (0.36–0.69), Pre 2 : 0.60 (0.42–0.77)) increased to almost perfect agreement (Post 1 : 0.74 (0.57–0.89), Post 2 : 0.90 (0.79–1)) post-standard. The intra-rater agreement of 2/5 participants was almost perfect post-standard, while it remained within substantial levels for the others. A linear mixed model ( χ 2 (3) = 8.85, p = 0.03) illustrated a slight increase in dashboard usability after incorporating the standard (Pre 1 : 84.09, Pre 2 : 81.36; Post 1 : 87.73, Post 2 : 87.27). Overall, the results highlighted that the subjective standard enhanced reliability of practitioner agreement for the selected decision.
Standards are pivotal for generating the evidence required to manage players in professional sport environments like rugby union. Resultantly, using a three-step qualitative approach, this study aimed to formulate a consensus as a subjective standard for evidence generation pertaining to player management. The consensus statement intended to identify evidence on peaks/troughs in player external training loads using Global Positioning System (GPS)-based information in the High-Performance Unit (HPU) of a Gallagher Premiership rugby union club. Initially, a systematic review adhering to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) framework was conducted to unravel the factors considered (literature-based cues) when identifying peaks/troughs in player external training loads using GPS information. Next, thematic analysis conducted on the data obtained from 7 semi-structured interviews with HPU staff highlighted that they consider 6 factors with 38 elements (practitioner-based cues) during player external training load management. Thereafter, guided by the Appraisal of Guidelines for Research and Evaluation (AGREE) II instrument and by utilising selected elements representing 4/6 factors (healthy player, GPS information, longitudinal durations and practitioner judgements on information), a consensus among practitioners for identifying peaks/troughs in player external training loads was developed with the participation of five HPU members using the nominal group technique (NGT). Practitioners reached an agreement with regard to 12 indicators to subjectively identify peaks/troughs in player external training loads within the considered environment.
Aim & Research Question Davis et al. 2019 proposed six visible signs (VS) of concussion. Our study evaluated frequency of VS after head impacts, how often individuals with VS underwent a diagnostic evaluation, and correlated head impact magnitudes with appearance of VS. Design Retrospective Observational Cohort/Setting Military service members in training and athletics, as well as and civilian athletes. Participants 542 military service members and 1502 civilians. Independent Variables On-site report from Commander or video confirmation rated by an expert. Outcome Measures VS of concussion, concussion assessment/removal from activity, impact monitoring mouthguard data. Main Results 177,800 impacts across 10900 subject-days of participation were collected. Most impacts (64%) were in daily living ranges (e.g. <10g acceleration and <1J workload). VS were seen in a small number of impacts (n=426, 0.2%). In VS impacts, 82% were not assessed and 86% continued participation. The median VS impact magnitudes (55g, 30J) versus non-VS impacts (8g, 1J) were significantly different (p<0.001 for both parameters). Conclusions This work supports the importance of identification of visible signs of suspected concussion, triggering appropriate diagnostic evaluation advocated by the 5th International Conference on Concussion. It appears that while potentially concussive events are uncommon, VS are often unseen. When VS did occur in our cohort they were frequently not assessed. Impacts with VS of concussion had significantly greater measured force than those without. These data suggest a well-calibrated impact monitoring system may help capture high magnitude impacts that could lead to VS.
Background The Head Injury Assessment (HIA) improve the identification of suspected concussions at elite levels, although whether certain teams, playing positions or player disproportionately cause or receive HIA in rugby league is unknown. Objective Describe and compare differences in incidence rates for players to cause or receive head injury assessments (HIAs), by playing position, team, and tackler event roles (i.e., ball carrier vs. tackler) in professional male rugby league. Design Retrospective cohort study Setting Super League rugby league. Participants 481 players from 13 teams in 305 matches during 2021 and 2022 Super League seasons. Assessment of Risk Factors Generalised linear mixed-effects models were used to compare incidence rate ratios between player event roles (i.e., ball carrier vs. tackler), playing positions, teams, and individual players. Main Outcome Measurements Head Injury Assessment (HIA) incidence rates and Incidence Rate Ratios. Results Tacklers had significantly greater incidence (1.56 HIAs per 1,000 tackles [95% confidence interval [CI] = 1.49–1.63]) than ball carriers (1.04 [0.97–1.11]). Compared to the median position, fullbacks (1.96 [1.09–3.52]), wingers (1.47 [0.85–2.54]), and centres (1.34 [0.86–2.10]) had the highest incidence of receiving HIAs. Wingers (1.64 [0.95–2.86]), fullbacks (1.14 [0.55–2.33]), and props (0.96 [95% CI = 0.68–1.35]) had the highest incidence of causing HIAs when compared to the median position. In comparison to the median team and player, three teams and five individual players had exhibited a significantly greater incidence of receiving HIAs. Additionally, three different teams and six different individual players had a significantly greater incidence of causing HIAs. Conclusions Given some playing positions, teams, and individual players have a greater incidence of causing or receiving HIA, targeted prevention strategy are required to reduce HIAs. This would focus on fullbacks, wingers, centres, props, and individual teams (3/13 causing and 3/13 receiving), and players (5/481 causing and 6/481 receiving).
The study aimed to establish the test-retest reliability of detrended fluctuation analysis of heart rate variability (DFA-alpha 1) based exercise intensity thresholds, assess its agreement with ventilatory- and lactate-derived thresholds and the moderating effect of sex and cardiorespiratory fitness (CRF) on the agreement. Intensity thresholds for thirty-seven participants (17 females) based on blood lactate (LT1/LT2), gas-exchange (VT1/VT2) and DFA-alpha 1 (alpha Th-1/alpha Th-2) were assessed. Heart rate (HR) at alpha Th-1 and alpha Th-2 showed good test-retest reliability (coefficient of variation [CV] < 6%), and moderate to high agreement with LTs (r = 0.40 - 0.57) and VTs (r = 0.61 - 0.66) respectively. Mixed effects models indicated bias magnitude depended on CRF, with DFA-alpha 1 overestimating thresholds versus VTs for lower fitness levels (speed at VT1 <8.5 km & sdot;hr(-1)), while underestimating for higher fitness levels (speed at VT2 >15 km & sdot;hr(-1); VO2max >55 mLkg(-1)min(-1)). Controlling for CRF, sex significantly affected bias magnitude only at first threshold, with males having higher mean bias (+2.41 bpm) than females (-1.26 bpm). DFA-alpha 1 thresholds are practical and reliable intensity measures, however it is unclear if they accurately represent LTs/VTs from the observed limits of agreement and unexplained variance. To optimise DFA-alpha 1 threshold estimation across different populations, bias should be corrected based on sex and CRF.
The cluster analysis of elite rugby league players identified groups of distinct playing positions, that can be referred to as broad positional groups. However, the identified positional groups were based on traditional indicators (physical and technical-tactical) that provided no information about the exact match-based movement activities that led to such similarity grouping and the classification of elite rugby league players into these broad positional groups remains unexplored. Hence, this study finds the best model to classify elite rugby league players into positional groups, using data characterised by movement patterns to uncover the similar movement activities of distinct playing positions within a positional group. Key movement patterns for the positional group classification and differences between the groups were also investigated. A total of 18,173 unique movement patterns were derived from 422 players' GPS data across the 2019 and 2020 seasons, where only 36 were identified as key patterns. The highest classification accuracy of 77.58% using all unique patterns and 74.5% accuracy using the key patterns was achieved, outperforming studies that used traditional indicators. Further analyses, based on key patterns revealed differences between forwards and backs. These findings establish movement patterns as viable indicators to classify rugby league players into positional groups, enabling coaches and trainers to develop position-specific training programs that cater to the unique physical demands of each position, leading to better player development and team performance. Movement patterns are therefore recommended as an alternative approach to quantifying players' external loads and obtaining granular information.