Stephan van der Zwaard

Aims: : Sport-type classification frameworks recommended by European (ESC) and American (AHA/ACC) guidelines are used to contextualize cardiac remodelling in athletes but remain unvalidated against cardiac magnetic resonance (CMR) phenotypes. We aimed to evaluate whether sport classifications predict CMR-determined cardiac remodelling phenotypes in elite athletes. Methods and Results: : We included 1,024 individuals: 775 elite athletes (310 women, 48 disciplines) and 249 controls. Athletes were classified by the ESC (four-category) and AHA/ACC (static–dynamic, four-quadrant) frameworks. CMR metrics were transformed to sex-specific z-scores (remodelling: z≥1.645). Because no CMR convention defines geometric phenotypes in athletes, the most strongly associated CMR metric was used to define each framework's assumed phenotype. Performance was evaluated at population (R2) and individual (balanced accuracy) levels. ESC categories were most strongly associated with RVEDV (R2=0.170); RV remodelling prevalence increased from skill to endurance (8.3%–62.2%; OR 3.82 per category; P<0.001). The AHA/ACC static component was most strongly associated with LV mass (R2=0.077) and the dynamic component with RVEDV (R2=0.072). Individual-level discrimination was poor: balanced accuracy was 34.2% (ESC) and 37.7% (AHA/ACC) (vs 25% chance). LV mass and RVEDV showed coupled increases (R2=0.23), with 67% of athletes demonstrating concordant patterns rather than discrete static or dynamic phenotypes. Misclassification was most pronounced for predominantly high-static athletes, of whom 73.8% showed normal geometry rather than expected LV hypertrophy (17.5%). Conclusions: : ESC and AHA/ACC sport classifications capture population-level remodelling but cannot reliably be used to predict individual cardiac phenotypes. In individual athletes interpretation should prioritise observed mass-volume relationships and the overall clinical context rather than sport type alone.

Sprint and endurance capacities seem to be mutually exclusive or at least at odds with each other. However, this relationship has not been investigated in soccer, which appeals to both well-developed sprint and endurance capacities. This study explores the potential of machine learning to identify soccer players based on their unique combinations of sprint and endurance capacities and sprint and endurance match-specific running performance. In this context, the relationships between sprint and endurance capacities and between physical capacities and match-specific running performance are examined in detail. Match-specific running data were collected from 31 young elite male soccer players over two consecutive seasons. Additionally, these participants underwent exercise testing, consisting of a 20-meter sprint test and an incremental treadmill test to measure maximal oxygen uptake (V̇O2max). Subgroups were identified using k-means clustering and subgroup discovery, based on players’ sprint and endurance capacities, sprint and endurance match-specific running performance, and playing position. Three distinct subgroups were identified using machine learning: players with high sprint capacity and sprinted meters (n = 4), players with high endurance capacity and meters ran at moderate and high intensities (n = 6), and players without high physical capacities or matching match-specific running performance (n = 14). Across all players, there was no significant relationship between 20-meter sprint speed and normalized V̇O2max (R2 = 0.085, P = 0.17), although 20-meter sprint speed was positively related to average match sprint distance (R2 = 0.168, P = 0.03) and normalized V̇O2max to average match distance at moderate and high intensities (R2 = 0.151, P = 0.04). In young elite soccer players, sprint and endurance capacities show positive, moderate, relationships with corresponding match-specific running performance, but those capacities do not appear to be mutually exclusive or opposing. Clustering allows for identification of players who may benefit from alternative strategic roles during matches, are at risk of overuse, or could benefit from individualized training. This method can assist coaches in designing tailored training programs and optimizing overall match strategy.

Soccer players are frequently categorized by playing positions, both in the scientific literature and in practice. However, the utility of this approach in evaluating physical match performance and optimizing physical training programs remains unclear. This study compares the effectiveness of categorizing soccer players by their playing position versus using unsupervised machine learning based on match-specific running performance. Match-specific running data were collected from 40 young elite male soccer players over two seasons. Thirty-one of these players completed a 20-meter sprint test and a maximal incremental treadmill test to measure maximal oxygen uptake. Players were categorized both by playing position and by subgroups derived through k-means clustering based on match-specific running performance. Differences in sprint capacity, endurance capacity, and match-specific running performance were compared between and within playing positions, as well as between and within clusters. The two categorization methods were further compared for variance within subgroups and standardized differences between subgroups for total distance (TD), low-intensity running (LIR), moderate-intensity running (MIR), high-intensity running (HIR), and sprint distance during matches. Match-specific running performance differed between playing positions, despite notable inter-individual differences in running intensities within playing positions. Clustering based on match-specific running performance revealed less variance within groups (TD: P = 0.049, LIR: P = 0.032, HIR: P = 0.033) and larger standardized differences between groups (LIR: P = 0.037, MIR: P = 0.041, HIR: P = 0.035, Sprint: P = 0.018) compared to grouping by playing position. Moreover, 20-meter sprint speed differed between the sprint and high intensity endurance clusters (25.22 vs 23.75 km/h, P = 0.012), but not between playing positions. Using unsupervised machine learning to categorize soccer players improves the identification of player groups with similar match-specific running performance, thereby supporting performance evaluation and contributing to the optimization of physical training.

Tandem cycling is a Paralympic discipline where two cyclists share one bicycle and requires synchronization and shared effort. In this study, we investigated how individual performance on a solo bicycle compares to tandem cycling. Twelve trained cyclists, that were unfamiliar with tandem cycling, completed a submaximal cycling test, a 30 s Wingate test, and a 10 min time trial under solo and tandem conditions. Performance metrics included gross efficiency, maximal power output, mean power output, heart rate, and cadence. The results showed that mean power output during the 10 min time trial was significantly lower on a tandem than on a solo bicycle (257 ± 66 W vs. 282 ± 80 W and p = 0.004), whereas cadence was higher (102.0 ± 7.7 rpm vs. 93.6 ± 4.7 rpm and p = 0.002). Gross efficiency, maximal power output, mean power output over 30 s, and heart rate over 10 min did not differ significantly between solo and tandem cycling. These findings suggest that although tandem cycling influences cadence and power, it does not reduce efficiency or short‐duration maximal performance. Given the differences in power output and cadence, coaches and athletes should carefully interpret solo cycling performance metrics when applied to tandem racing and training, especially in novice tandem cyclists. Future research should investigate the role of power output matching and tandem experience, and further explore the interaction between pilot and stoker to optimize performance outcomes.

Background Performance optimization is a major goal in sports science. However, this remains difficult due to the small samples and large individual variation in physiology and training adaptations. Machine learning (ML) solutions seem promising, but have not been tested for their capability to predict performance in this setting. The aim of this study was to predict 4-km cycling performance following a 12-week training intervention based on ML models with predictors from physiological profiling, individual training load and well-being, and to retrieve the most important predictors. Specific techniques were applied to reduce the risk of overfitting. Results Twenty-seven recreational cyclists completed the 4-km time trial with a mean power output of 4.1 ± 0.7 W/kg. Changes in time-trial performance after training were not different between moderate-intensity endurance training (n = 6), polarised endurance training (n = 8), concurrent polarised with concentric strength training (n = 7) and concurrent polarised with eccentric strength training (n = 6) groups (P > 0.05), but included substantial inter-individual differences. ML models predicted cycling performance with excellent model performance on unseen data before (R² = 0.923, mean absolute error (MAE) = 0.183 W/kg using a generalized linear model) and after training (R² = 0.758, MAE = 0.338 W/kg using a generalized linear model). Absolute changes in performance were more difficult to predict (R² = 0.483, MAE = 0.191 W/kg using a random forest model). Important predictors included power at V̇O_2max, performance V̇O₂, ventilatory thresholds and efficiency, but also parameters related to body composition, training impulse, sleep, sickness and well-being. Conclusion ML models allow accurate predictions of cycling performance based on physiological profiling, individual training load and well-being during a 12-week training intervention, even using small sample sizes, although changes in cycling performance were more difficult to predict.

Tandem cycling is a paralympic discipline, in which two cyclists ride on one tandem bicycle. Their performance can be improved by minimizing power losses. This study aimed to quantify power loss due to the chain drive of a tandem bicycle and influence of power input, location of power input, and rear chain wheel size. Power loss was determined by the difference between power input applied on the cranks (front or back rider) and power output measured at the rear wheel. Power output values were set from 100 to 400 W, with 50 W increments, and using two gear ratios, 53-11 and 53-13. Power input was generated at the back crank (with only the primary chain—solo bicycle—and with primary and secondary chain attached) and at the front crank (with primary and secondary chain attached). The power loss for the solo bicycle was 2.1% (± 1.5%). A significantly larger power loss was found for the tandem in the back (3.7 ± 2.4%, p < 0.001) and front positions (3.0 ± 1.8%, p < 0.001), with marginal differences between positions at higher power output. Power loss for the tandem was lower with gear 53-13 (2.9% ± 1.7%) compared to 53-11 (4.0 ± 2.8%, p < 0.001, effect size is medium). Therefore, findings suggest that back and front riders experience similar power losses due to the chain drive, but more than on a solo bicycle. Tandem cyclists can reduce their power loss in the chain drive by selecting larger gear ratios. Coaches may consider these findings for selecting and coaching their tandem cyclists.

Background: Training load is typically described in terms of internal and external load. Investigating the coupling of internal and external training load is relevant to many sports. Here, continuous kernel-density estimation (KDE) may be a valuable tool to capture and visualize this coupling. Aim: Using training load data in speed skating, we evaluated how well bivariate KDE plots describe the coupling of internal and external load and differentiate between specific training sessions, compared to training impulse scores or intensity distribution into training zones. Methods: On-ice training sessions of 18 young (sub)elite speed skaters were monitored for velocity and heart rate during 2 consecutive seasons. Training session types were obtained from the coach’s training scheme, including endurance, interval, tempo, and sprint sessions. Differences in training load between session types were assessed using Kruskal–Wallis or Kolmogorov–Smirnov tests for training impulse and KDE scores, respectively. Results: Training impulse scores were not different between training session types, except for extensive endurance sessions. However, all training session types differed when comparing KDEs for heart rate and velocity (both P < .001). In addition, 2D KDE plots of heart rate and velocity provide detailed insights into the (subtle differences in) coupling of internal and external training load that could not be obtained by 2D plots using training zones. Conclusion: 2D KDE plots provide a valuable tool to visualize and inform coaches on the (subtle differences in) coupling of internal and external training load for training sessions. This will help coaches design better training schemes aiming at desired training adaptations.

Myoglobin is essential for oxygen transport to the muscle fibers. However, measurements of myoglobin (Mb) protein concentrations within individual human muscle fibers are scarce. Recent observations have revealed surprisingly low Mb concentrations in elite cyclists, however it remains unclear whether this relates to Mb translation, transcription and/or myonuclear content. The aim was to compare Mb concentration, Mb messenger RNA (mRNA) expression levels and myonuclear content within muscle fibers of these elite cyclists with those of physically-active controls. Muscle biopsies were obtained from m. vastus lateralis in 29 cyclists and 20 physically-active subjects. Mb concentration was determined by peroxidase staining for both type I and type II fibers, Mb mRNA expression level was determined by quantitative PCR and myonuclear domain size (MDS) was obtained by immunofluorescence staining. Average Mb concentrations (mean ± SD: 0.38 ± 0.04 mM vs. 0.48 ± 0.19 mM; P = 0.014) and Mb mRNA expression levels (0.067 ± 0.019 vs. 0.088 ± 0.027; P = 0.002) were lower in cyclists compared to controls. In contrast, MDS and total RNA per mg muscle were not different between groups. Interestingly, in cyclists compared to controls, Mb concentration was only lower for type I fibers (P < 0.001), but not for type II fibers (P > 0.05). In conclusion, the lower Mb concentration in muscle fibers of elite cyclists is partly explained by lower Mb mRNA expression levels per myonucleus and not by a lower myonuclear content. It remains to be determined whether cyclists may benefit from strategies that upregulate Mb mRNA expression levels, particularly in type I fibers, to enhance their oxygen supply.