Influence of the Mechanical Properties of Third-Generation Artificial Turf Systems on Soccer Players’ Physiological and Physical Performance and Their Perceptions

The aim of this research was to evaluate the influence of the mechanical properties of artificial turf systems on soccer players’ performance. A battery of perceptive physiological and physical tests were developed on four different structural systems of artificial turf (System 1: Compacted gravel sub-base without elastic layer; System 2: Compacted gravel sub-base with elastic layer; System 3: Asphalt sub-base without elastic layer; System 4: Asphalt sub-base with elastic layer). The sample was composed of 18 soccer players (22.44±1.72 years) who typically train and compete on artificial turf. The artificial turf system with less rotational traction (S3) showed higher total time in the Repeated Sprint Ability test in comparison to the systems with intermediate values (49.46±1.75 s vs 47.55±1.82 s (S1) and 47.85±1.59 s (S2); p<0.001). The performance in jumping tests (countermovement jump and squat jump) and ball kicking to goal decreased after the RSA test in all surfaces assessed (p<0.05), since the artificial turf system did not affect performance deterioration (p>0.05). The physiological load was similar in all four artificial turf systems. However, players felt more comfortable on the harder and more rigid system (S4; visual analogue scale = 70.83±14.28) than on the softer artificial turf system (S2; visual analogue scale = 54.24±19.63). The lineal regression analysis revealed a significant influence of the mechanical properties of the surface of 16.5%, 15.8% and 7.1% on the mean time of the sprint, the best sprint time and the maximum mean speed in the RSA test respectively. Results suggest a mechanical heterogeneity between the systems of artificial turf which generate differences in the physical performance and in the soccer players’ perceptions.     

Combining small-sided soccer games and running-based methods: A systematic review

Small-sided games (SSGs) are often used in soccer to produce acute physiological and physical responses, while a tactical/technical stimulus is also employed. However, due to some limitations of SSGs, researchers have been testing this method combined with running-based training methods. This systematic review was conducted to assess the effects of combined SSG and running-based methods on soccer players’ acute responses and adaptations after training interventions. A systematic review of Web of Science, PubMed, Cochrane Library, Scopus, and SPORTDiscus databases was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The database search initially identified 782 titles. From those, five articles were deemed eligible for the systematic review. The five included studies presented data from training load, reporting inconsistent greater values in combined SSG and running-based methods when compared to SSG-only formats. Considering the adaptations, studies comparing combined SSG and running-based methods with SSG-only methods revealed inconsistent differences in terms of the effects on aerobic performance and sprinting. Combining SSG and running-based methods can increase the acute mechanical load and high-intense running stimuli in players when compared to interventions that use only SSGs. However, the adaptations promoted by both methods are similar, and the differences are unclear. The order of combination (SSG and running-based method) does not seem to impact players’ adaptations; however, the frequency of sessions did have a meaningful impact.     

A Method for Using Player Tracking Data in Basketball to Learn Player Skills and Predict Team Performance

Player tracking data represents a revolutionary new data source for basketball analysis, in which essentially every aspect of a player’s performance is tracked and can be analyzed numerically. We suggest a way by which this data set, when coupled with a network-style model of the offense that relates players’ skills to the team’s success at running different plays, can be used to automatically learn players’ skills and predict the performance of untested 5-man lineups in a way that accounts for the interaction between players’ respective skill sets. After developing a general analysis procedure, we present as an example a specific implementation of our method using a simplified network model. While player tracking data is not yet available in the public domain, we evaluate our model using simulated data and show that player skills can be accurately inferred by a simple statistical inference scheme. Finally, we use the model to analyze games from the 2011 playoff series between the Memphis Grizzlies and the Oklahoma City Thunder and we show that, even with a very limited data set, the model can consistently describe a player’s interactions with a given lineup based only on his performance with a different lineup.     

Adaptive search space pruning in complex strategic problems

People have limited computational resources, yet they make complex strategic decisions over enormous spaces of possibilities. How do people efficiently search spaces with combinatorially branching paths? Here, we study players’ search strategies for a winning move in a “k-in-a-row” game. We find that players use scoring strategies to prune the search space and augment this pruning by a “shutter” heuristic that focuses the search on the paths emanating from their previous move. This strong pruning has its costs—both computational simulations and behavioral data indicate that the shutter size is correlated with players’ blindness to their opponent’s winning moves. However, simulations of the search while varying the shutter size, complexity levels, noise levels, branching factor, and computational limitations indicate that despite its costs, a narrow shutter strategy is the dominant strategy for most of the parameter space. Finally, we show that in the presence of computational limitations, the shutter heuristic enhances the performance of deep learning networks in these end-game scenarios. Together, our findings suggest a novel adaptive heuristic that benefits search in a vast space of possibilities of a strategic game.     

How weekly monitoring variables influence players’ and teams’ match performance in elite futsal players

This study aimed to investigate how weekly training load constrains the performance of players and teams in official futsal competitions. Data from a professional male team were collected during two seasons (46 weeks). The applied monitoring system analysed the training load (as measured by session perceived exertion, sRPE), the total recovery status (TQR), the well-being score (WBs) and the variability of neuromuscular performance during each week (CMJ-cv). In addition, the performance was assessed for all the matches. A path analysis model was performed to test the associations across variables. Results from the path analysis model revealed that it explains 31% of the teams’ performance. In general, the results show that previous team performance has no significant effects on the training week. A significant negative relationship was found between CMJ-cv and match performance (β = -.34; CI95% -.359 to -.070), as well as a significant negative relationship between players’ match performance and the team’s match performance (β = -.55; CI95% -.292 to .740). Regarding indirect effects, only a negative association between CMJ-cv and team match performance via players’ match performance (β = -.19; CI95% -.342 to -.049) was identified. The small variation of the weekly CMJ (CMJ-cv) seems to be a key variable to monitor and explain both player and team performance. Based on this model, and only looking at the physical variables, it was possible to explain 31% of the team’s performance. Longitudinal and multi-team studies should be conducted to integrate other technical, tactical and psychological variables that allow the level of understanding of players’ and teams’ performance to be improved.     

Why Robots Should Be Social: Enhancing Machine Learning through Social Human-Robot Interaction

Social learning is a powerful method for cultural propagation of knowledge and skills relying on a complex interplay of learning strategies, social ecology and the human propensity for both learning and tutoring. Social learning has the potential to be an equally potent learning strategy for artificial systems and robots in specific. However, given the complexity and unstructured nature of social learning, implementing social machine learning proves to be a challenging problem. We study one particular aspect of social machine learning: that of offering social cues during the learning interaction. Specifically, we study whether people are sensitive to social cues offered by a learning robot, in a similar way to children’s social bids for tutoring. We use a child-like social robot and a task in which the robot has to learn the meaning of words. For this a simple turn-based interaction is used, based on language games. Two conditions are tested: one in which the robot uses social means to invite a human teacher to provide information based on what the robot requires to fill gaps in its knowledge (i.e. expression of a learning preference); the other in which the robot does not provide social cues to communicate a learning preference. We observe that conveying a learning preference through the use of social cues results in better and faster learning by the robot. People also seem to form a “mental model” of the robot, tailoring the tutoring to the robot’s performance as opposed to using simply random teaching. In addition, the social learning shows a clear gender effect with female participants being responsive to the robot’s bids, while male teachers appear to be less receptive. This work shows how additional social cues in social machine learning can result in people offering better quality learning input to artificial systems, resulting in improved learning performance.     

Predicting ratings of perceived exertion in youth soccer using decision tree models

The purpose of this study was to determine the effectiveness of white-box decision tree models (DTM) for predicting the rating of perceived exertion (RPE). The second aim was to examine the relationship between RPE and external measures of intensity in youth soccer training at the group and individual level. Training load data from 18 youth soccer players were collected during an in-season competition period. A total of 804 training observations were undertaken, with a total of 43 ± 17 sessions per player (range 12–76). External measures of intensity were determined using a 10 Hz GPS and included total distance (TD, m/min), high-speed running distance (HSR, m/min), PlayerLoad (PL, n/min), impacts (n/min), distance in acceleration/deceleration (TD ACC/TD DEC, m/min) and the number of accelerations/decelerations (ACC/DEC, n/min). Data were analysed with decision tree models. Global and individualized models were constructed. Aggregated importance revealed HSR as the strongest predictor of RPE with relative importance of 0.61. HSR was the most important factor in predicting RPE for half of the players. The prediction error (root mean square error [RMSE] 0.755 ± 0.014) for the individualized models was lower compared to the population model (RMSE 1.621 ± 0.001). The findings demonstrate that individual models should be used for the assessment of players’ response to external load. Furthermore, the study demonstrates that DTM provide straightforward interpretation, with the possibility of visualization. This method can be used to prescribe daily training loads on the basis of predicted, desired player responses (exertion).     

Can Perceptuo-Motor Skills Assessment Outcomes in Young Table Tennis Players (7–11 years) Predict Future Competition Participation and Performance? An Observational Prospective Study

Forecasting future performance in youth table tennis players based on current performance is complex due to, among other things, differences between youth players in growth, development, maturity, context and table tennis experience. Talent development programmes might benefit from an assessment of underlying perceptuo-motor skills for table tennis, which is hypothesized to determine the players’ potential concerning the perceptuo-motor domain. The Dutch perceptuo-motor skills assessment intends to measure the perceptuo-motor potential for table tennis in youth players by assessing the underlying skills crucial for developing technical and tactical qualities. Untrained perceptuo-motor tasks are used as these are suggested to represent a player’s future potential better than specific sport skills themselves as the latter depend on exposure to the sport itself. This study evaluated the value of the perceptuo-motor skills assessment for a talent developmental programme by evaluating its predictive validity for competition participation and performance in 48 young table tennis players (7–11 years). Players were tested on their perceptuo-motor skills once during a regional talent day, and the subsequent competition results were recorded half-yearly over a period of 2.5 years. Logistic regression analysis showed that test scores did not predict future competition participation (p >0.05). Yet, the Generalized Estimating Equations analysis, including the test items ‘aiming at target’, ‘throwing a ball’, and ‘eye-hand coordination’ in the best fitting model, revealed that the outcomes of the perceptuo-motor skills assessment were significant predictors for future competition results (R² = 51%). Since the test age influences the perceptuo-motor skills assessment’s outcome, another multivariable model was proposed including test age as a covariate (R² = 53%). This evaluation demonstrates promising prospects for the perceptuo-motor skills assessment to be included in a talent development programme. Future studies are needed to clarify the predictive value in a larger sample of youth competition players over a longer period in time.     

Reducing the Socio-Economic Status Achievement Gap at University by Promoting Mastery-Oriented Assessment

In spite of official intentions to reduce inequalities at University, students’ socio-economic status (SES) is still a major determinant of academic success. The literature on the dual function of University suggests that University serves not only an educational function (i.e., to improve students’ learning), but also a selection function (i.e., to compare people, and orient them towards different positions in society). Because current assessment practices focus on the selection more than on the educational function, their characteristics fit better with norms and values shared by dominant high-status groups and may favour high-SES students over low-SES students in terms of performances. A focus on the educational function (i.e., mastery goals), instead, may support low-SES students’ achievement, but empirical evidence is currently lacking. The present research set out to provide such evidence and tested, in two field studies and a randomised field experiment, the hypothesis that focusing on University’s educational function rather than on its selection function may reduce the SES achievement gap. Results showed that a focus on learning, mastery-oriented goals in the assessment process reduced the SES achievement gap at University. For the first time, empirical data support the idea that low-SES students can perform as well as high-SES students if they are led to understand assessment as part of the learning process, a way to reach mastery goals, rather than as a way to compare students to each other and select the best of them, resulting in performance goals. This research thus provides a theoretical framework to understand the differential effects of assessment on the achievement of high and low-SES students, and paves the way toward the implementation of novel, theory-driven interventions to reduce the SES-based achievement gap at University.     

Machine learning-based risk factor analysis and prevalence prediction of intestinal parasitic infections using epidemiological survey data

BackgroundPrevious epidemiological studies have examined the prevalence and risk factors for a variety of parasitic illnesses, including protozoan and soil-transmitted helminth (STH, e.g., hookworms and roundworms) infections. Despite advancements in machine learning for data analysis, the majority of these studies use traditional logistic regression to identify significant risk factors.MethodsIn this study, we used data from a survey of 54 risk factors for intestinal parasitosis in 954 Ethiopian school children. We investigated whether machine learning approaches can supplement traditional logistic regression in identifying intestinal parasite infection risk factors. We used feature selection methods such as InfoGain (IG), ReliefF (ReF), Joint Mutual Information (JMI), and Minimum Redundancy Maximum Relevance (MRMR). Additionally, we predicted children’s parasitic infection status using classifiers such as Logistic Regression (LR), Support Vector Machines (SVM), Random Forests (RF) and XGBoost (XGB), and compared their accuracy and area under the receiver operating characteristic curve (AUROC) scores. For optimal model training, we performed tenfold cross-validation and tuned the classifier hyperparameters. We balanced our dataset using the Synthetic Minority Oversampling (SMOTE) method. Additionally, we used association rule learning to establish a link between risk factors and parasitic infections.Key findingsOur study demonstrated that machine learning could be used in conjunction with logistic regression. Using machine learning, we developed models that accurately predicted four parasitic infections: any parasitic infection at 79.9% accuracy, helminth infection at 84.9%, any STH infection at 95.9%, and protozoan infection at 94.2%. The Random Forests (RF) and Support Vector Machines (SVM) classifiers achieved the highest accuracy when top 20 risk factors were considered using Joint Mutual Information (JMI) or all features were used. The best predictors of infection were socioeconomic, demographic, and hematological characteristics.ConclusionsWe demonstrated that feature selection and association rule learning are useful strategies for detecting risk factors for parasite infection. Additionally, we showed that advanced classifiers might be utilized to predict children’s parasitic infection status. When combined with standard logistic regression models, machine learning techniques can identify novel risk factors and predict infection risk.     

Impact of elite soccer coaching change on team performance according to coach- and club-related variables

A coaching change is an extreme, but frequently occurring phenomenon in elite soccer with its impact on team success debatable. The aim of the current study was twofold: (i) to compare team’s performance when coached by new and old coaches; and (ii) to investigate the impact of a coaching change on team’s performance according to coach- and club-related factors. All in-season coaching changes from the 2010–11 to 2017–18 seasons within the Spanish, French, English, German and Italian professional leagues were examined. Team performance was assessed as points awarded from match outcome over 1–20 matches prior to and following the coaching change. Four independent variables (coach’s experience, team’s budget, whether the coach had been an elite former player or not, and whether the coach was a novice or not) were included into linear regression modelling. The main results showed that team’s short-term performance was improved significantly with a change to a new coach with this impact declining in the longer term (> 10 matches). Specifically, the number of points (1.15–1.32 vs. 0.37–1.03, p < 0.05) and the moving average of points (1.19–1.31 vs. 0.37–1.04, p < 0.05) awarded per match were significantly greater after the coaching change. Further, the winning effect due to the new coach was independent of coach-related factors such as coaching experience or the new coach being a former elite player. A critical organisational decision to change coaches may provide an essential stimulus for future team success in elite soccer.     

The dominant leg is more likely to get injured in soccer players: systematic review and meta-analysis

In soccer (football), dominant limb kicking produces higher ball velocity and is used with greater frequency than the non-dominant limb. It is unclear whether limb dominance has an effect on injury incidence. The purpose of this systematic review with meta-analysis is to examine the relationship between limb dominance and soccer injuries. Studies were identified from four online databases according to PRISMA guidelines to identify studies of soccer players that reported lower extremity injuries by limb dominance. Relevant studies were assessed for inclusion and retained. Data from retained studies underwent meta-analyses to determine relative risk of dominant versus non-dominant limb injuries using random-effects models. Seventy-four studies were included, with 36 of them eligible for meta-analysis. For prospective lower extremity injury studies, soccer players demonstrated a 1.6 times greater risk of injury to the dominant limb (95% CI [1.3–1.8]). Grouped by injury location, hamstring (RR 1.3 [95% CI 1.1–1.4]) and hip/groin (RR 1.9 [95% CI 1.3–2.7]) injuries were more likely to occur to the dominant limb. Greater risk of injury was present in the dominant limb across playing levels (amateurs RR 2.6 [95% CI 2.1–3.2]; youths RR 1.5 [95% CI 1.26–1.67]; professionals RR 1.3 [95% CI 1.14–1.46]). Both males (RR 1.5 [95% CI 1.33–1.68)] and females (RR 1.5 [95% CI 1.14–1.89]) were more likely to sustain injuries to the dominant limb. Future studies investigating soccer injury should adjust for this confounding factor by using consistent methods for assigning limb dominance and tracking use of the dominant versus non-dominant limb.     

The Effects of Individualized Resistance Strength Programs on Knee Muscular Imbalances in Junior Elite Soccer Players

The purpose of this study was to investigate the effects of a resistance training program on the muscular strength of soccer players’ knees that initially presented unilateral and bilateral differences. For this study, a team of 24 male well-trained junior soccer players was divided into two strength program training groups: a Resistance Training Control Group (RTCG) composed of 10 players that did not have muscular imbalances and a Resistance Training Experimental Group (RTEG) composed of 14 players that had muscular imbalances. All players followed a resistance training program for six weeks, two times per week, during the transition period. The program of individualized strength training consisted of two parts. The first part, which was identical in terms of the choice of training loads, was intended for both training groups and contained two series of exercises including upper and lower body exercises. The second part of the program was intended only for RTEG and consisted of two additional series for the groups of muscles that had identified unilateral and bilateral differences. The applied program showed various directions in the isokinetic profile of changes. In the case of RTCG, the adaptations related mainly to the quadriceps muscle (the peak torque (PT) change for the dominant leg was statistically significant (p < 0.05)). There were statistically significant changes in RTEG (p < 0.05) related to PT for the hamstrings in both legs, which in turn resulted in an increase in the conventional hamstring/quadriceps ratio (H/Q). It is interesting that the statistically significant (p < 0.05) changes were noted only for the dominant leg. No statistically significant changes in bilateral differences (BD) were noted in either group. These results indicate that individualized resistance training programs could provide additional benefits to traditional strength training protocols to improve muscular imbalances in post-adolescent soccer players.     

Machine learning liver-injuring drug interactions with non-steroidal anti-inflammatory drugs (NSAIDs) from a retrospective electronic health record (EHR) cohort

Drug-drug interactions account for up to 30% of adverse drug reactions. Increasing prevalence of electronic health records (EHRs) offers a unique opportunity to build machine learning algorithms to identify drug-drug interactions that drive adverse events. In this study, we investigated hospitalizations’ data to study drug interactions with non-steroidal anti-inflammatory drugs (NSAIDS) that result in drug-induced liver injury (DILI). We propose a logistic regression based machine learning algorithm that unearths several known interactions from an EHR dataset of about 400,000 hospitalization. Our proposed modeling framework is successful in detecting 87.5% of the positive controls, which are defined by drugs known to interact with diclofenac causing an increased risk of DILI, and correctly ranks aggregate risk of DILI for eight commonly prescribed NSAIDs. We found that our modeling framework is particularly successful in inferring associations of drug-drug interactions from relatively small EHR datasets. Furthermore, we have identified a novel and potentially hepatotoxic interaction that might occur during concomitant use of meloxicam and esomeprazole, which are commonly prescribed together to allay NSAID-induced gastrointestinal (GI) bleeding. Empirically, we validate our approach against prior methods for signal detection on EHR datasets, in which our proposed approach outperforms all the compared methods across most metrics, such as area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC).     

Left-Handedness in Professional and Amateur Tennis

Negative frequency-dependent effects rather than innate predispositions may provide left-handers with an advantage in one-on-one fighting situations. Support mainly comes from cross-sectional studies which found significantly enhanced left-hander frequencies among elite athletes exclusively in interactive sports such as baseball, cricket, fencing and tennis. Since professional athletes’ training regimes continuously improve, however, an important unsolved question is whether the left-handers’ advantage in individual sports like tennis persists over time. To this end, we longitudinally tracked left-hander frequencies in year-end world rankings (men: 1973–2011, ladies: 1975–2011) and at Grand Slam tournaments (1968–2011) in male and female tennis professionals. Here we show that the positive impact of left-handed performance on high achievement in elite tennis was moderate and decreased in male professionals over time and was almost absent in female professionals. For both sexes, left-hander frequencies among year-end top 10 players linearly decreased over the period considered. Moreover, left-handedness was, however, no longer seems associated with higher probability of attaining high year-end world ranking position in male professionals. In contrast, cross-sectional data on left-hander frequencies in male and female amateur players suggest that a left-handers’ advantage may still occur on lower performance levels. Collectively, our data is in accordance with the frequency-dependent hypothesis since reduced experience with left-handers in tennis is likely to be compensated by players’ professionalism.     

Validity and Reliability of New Agility Test among Elite and Subelite under 14-Soccer Players

Background

Agility is a determinant component in soccer performance. This study aimed to evaluate the reliability and sensitivity of a “Modified Illinois change of direction test” (MICODT) in ninety-five U-14 soccer players.

Methods

A total of 95 U-14 soccer players (mean ± SD: age: 13.61±1.04 years; body mass: 30.52±4.54 kg; height: 1.57±0.1 m) from a professional and semi-professional soccer academy, participated to this study. Sixty of them took part in reliability analysis and thirty-two in sensitivity analysis.

Results

The intraclass correlation coefficient (ICC) that aims to assess relative reliability of the MICODT was of 0.99, and its standard error of measurement (SEM) for absolute reliability was <5% (1.24%). The MICODT’s capacity to detect change is “good”, it’s SEM (0.10 s) was ≤ SWC (0.33 s). The MICODT is significantly correlated to the Illinois change of direction speed test (ICODT) (r = 0.77; p<0.0001). The ICODT’s MDC₉₅ (0.64 s) was twice about the MICODT’s MDC₉₅ (0.28 s), indicating that MICODT presents better ability to detect true changes than ICODT. The MICODT provided good sensitivity since elite U-14 soccer players were better than non-elite one on MICODT (p = 0.005; d_z = 1.01 [large]). This was supported by an area under the ROC curve of 0.77 (CI 95%, 0.59 to 0.89, p<0.0008). The difference observed in these two groups in ICODT was not statistically significant (p = 0.14; d_z = 0.51 [small]), showing poor discriminant ability.

Conclusion

MICODT can be considered as more suitable protocol for assessing agility performance level than ICODT in U-14 soccer players.     

Accelerometry-based variables in professional soccer players: comparisons between periods of the season and playing positions

The aim of this study was to provide reference data of variation in external training loads for weekly periods within the annual season. Specifically, we aimed to compare the weekly acute load, monotony, and training strain of accelerometry-based measures across a professional soccer season (pre-season, first and second halves of the season) according to players’ positions. Nineteen professional players were monitored daily for 45 weeks using an 18-Hz global positioning system to obtain measures of high metabolic load distance (HMLD), impacts, and high intensity accelerations and decelerations. Workload indices of acute load, training monotony, and training strain were calculated weekly for each of the measures. The HMLD had greater training strain values in the pre-season than in the first (p ≤ 0.001; d = 0.793) and second halves of the season (p ≤ 0.001; d = 0.858). Comparisons between playing positions showed that midfielders had the highest weekly acute load of HMLD (6901 arbitrary units [AU]), while central defenders had the lowest (4986 AU). The pre-season period was associated with the highest acute and strain load of HMLD and number of impacts, with a progressive decrease seen during the season. In conclusion, coaches should consider paying greater attention to variations in HMLD and impacts between periods of the season and between players to individualize training accordingly.     

Machine learning in human movement biomechanics: Best practices,common pitfalls,and new opportunities

Traditional laboratory experiments, rehabilitation clinics, and wearable sensors offer biomechanists a wealth of data on healthy and pathological movement. To harness the power of these data and make research more efficient, modern machine learning techniques are starting to complement traditional statistical tools. This survey summarizes the current usage of machine learning methods in human movement biomechanics and highlights best practices that will enable critical evaluation of the literature. We carried out a PubMed/Medline database search for original research articles that used machine learning to study movement biomechanics in patients with musculoskeletal and neuromuscular diseases. Most studies that met our inclusion criteria focused on classifying pathological movement, predicting risk of developing a disease, estimating the effect of an intervention, or automatically recognizing activities to facilitate out-of-clinic patient monitoring. We found that research studies build and evaluate models inconsistently, which motivated our discussion of best practices. We provide recommendations for training and evaluating machine learning models and discuss the potential of several underutilized approaches, such as deep learning, to generate new knowledge about human movement. We believe that cross-training biomechanists in data science and a cultural shift toward sharing of data and tools are essential to maximize the impact of biomechanics research.     

Players’ physical performance in LaLiga across the season: insights for competition continuation after COVID-19

Due to the COVID-19 outbreak, professional football players competing in LaLiga were confined at home for ~8 weeks and then they were allowed to train to prepare the first competitive match for 4 weeks. As the duration of summer break in the prior four seasons of LaLiga (from 2015-2016 to 2018-2019) was of similar length to the suspension of the championship due to COVID-19 (~12 weeks), we have analysed the running performance of teams competing in LaLiga in these four seasons to anticipate players’ physical performance after the resumption of the competition. The analysis includes the average running distance per game for each of the 38 matchdays that compose LaLiga. One-way ANOVA revealed that there was a main effect of the matchday on total running distance per match (p = 0.001), and in the distance covered between 14.0 and 20.9 km/h (p < 0.001), between 21.0 and 23.9 km/h (p < 0.001) and at above 24.0 km/h (p < 0.001). Overall, the post-hoc analysis revealed that the running patterns progressively increased during the first 8-10 matchdays and then reached a plateau which was significantly different to matchday-1 (p < 0.05). This analysis reveals that, in the prior four competitive seasons of LaLiga, players’ physical performance was lower at the beginning of the season and the teams needed approximately 8-10 matchdays to reach a steady state running performance. These data suggest that football players will progressively increase their performance across the 11 matchdays remaining to complete LaLiga.     

Improving the Spatial Prediction of Soil Organic Carbon Stocks in a Complex Tropical Mountain Landscape by Methodological Specifications in Machine Learning Approaches

Tropical forests are significant carbon sinks and their soils’ carbon storage potential is immense. However, little is known about the soil organic carbon (SOC) stocks of tropical mountain areas whose complex soil-landscape and difficult accessibility pose a challenge to spatial analysis. The choice of methodology for spatial prediction is of high importance to improve the expected poor model results in case of low predictor-response correlations. Four aspects were considered to improve model performance in predicting SOC stocks of the organic layer of a tropical mountain forest landscape: Different spatial predictor settings, predictor selection strategies, various machine learning algorithms and model tuning. Five machine learning algorithms: random forests, artificial neural networks, multivariate adaptive regression splines, boosted regression trees and support vector machines were trained and tuned to predict SOC stocks from predictors derived from a digital elevation model and satellite image. Topographical predictors were calculated with a GIS search radius of 45 to 615 m. Finally, three predictor selection strategies were applied to the total set of 236 predictors. All machine learning algorithms—including the model tuning and predictor selection—were compared via five repetitions of a tenfold cross-validation. The boosted regression tree algorithm resulted in the overall best model. SOC stocks ranged between 0.2 to 17.7 kg m^-2, displaying a huge variability with diffuse insolation and curvatures of different scale guiding the spatial pattern. Predictor selection and model tuning improved the models’ predictive performance in all five machine learning algorithms. The rather low number of selected predictors favours forward compared to backward selection procedures. Choosing predictors due to their indiviual performance was vanquished by the two procedures which accounted for predictor interaction.