Purpose: Staying injury free is a major factor for success in sports. Although injuries are difficult to forecast, novel technologies and data-science applications could provide important insights. Our purpose was to use machine learning for the prediction of injuries in runners, based on detailed training logs. Methods: Prediction of injuries was evaluated on a new data set of 74 high-level middle- and long-distance runners, over a period of 7 years. Two analytic approaches were applied. First, the training load from the previous 7 days was expressed as a time series, with each day’s training being described by 10 features. These features were a combination of objective data from a global positioning system watch (eg, duration, distance), together with subjective data about the exertion and success of the training. Second, a training week was summarized by 22 aggregate features, and a time window of 3 weeks before the injury was considered. Results: A predictive system based on bagged XGBoost machine-learning models resulted in receiver operating characteristic curves with average areas under the curves of 0.724 and 0.678 for the day and week approaches, respectively. The results of the day approach especially reflect a reasonably high probability that our system makes correct injury predictions. Conclusions: Our machine-learning-based approach predicts a sizable portion of the injuries, in particular when the model is based on training-load data in the days preceding an injury. Overall, these results demonstrate the possible merits of using machine learning to predict injuries and tailor training programs for athletes.
S. Sofie Lövdal, Ruud J.R. Den Hartigh, and George Azzopardi
Stephen S. Cheung
Lasse Ishøi, Kasper Krommes, Mathias F. Nielsen, Kasper B. Thornton, Per Hölmich, Per Aagaard, Juan J.J. Penalver, and Kristian Thorborg
Purpose: Increasing age, high quadriceps strength, and low hamstring muscle strength are associated with hamstring strain injury in soccer. The authors investigated the age-related variation in maximal hamstring and quadriceps strength in male elite soccer players from under-13 (U-13) to the senior level. Methods: A total of 125 elite soccer players were included from a Danish professional soccer club and associated youth academy (first tier; U-13, n = 19; U-14, n = 16; U-15, n = 19; U-17, n = 24; U-19, n = 17; and senior, n = 30). Maximal voluntary isometric force was assessed for the hamstrings at 15° knee joint angle and for the quadriceps at 60° knee joint angle (0° = full extension) using an external-fixated handheld dynamometer. Hamstring-to-quadriceps strength (H:Q) ratio and hamstring and quadriceps maximal voluntary isometric force levels were compared across age groups (U-13 to senior). Results: Senior players showed 18% to 26% lower H:Q ratio compared with all younger age groups (P ≤ .026). Specific H:Q ratios (mean [95% confidence interval]) were as follows: senior, 0.45 (0.42–0.48); U-19, 0.61 (0.55–0.66); U-17, 0.56 (0.51–0.60); U-15, 0.59 (0.54–0.64); U-14, 0.54 (0.50–0.59); and U-13, 0.57 (0.51–0.62). Hamstring strength increased from U-13 to U-19 with a significant drop from U-19 to the senior level (P = .048), whereas quadriceps strength increased gradually from U-13 to senior level. Conclusion: Elite senior soccer players demonstrate lower H:Q ratio compared with youth players, which is driven by lower hamstring strength at the senior level compared with the U-19 level combined with a higher quadriceps strength. This discrepancy in hamstring and quadriceps strength capacity may place senior-level players at increased risk of hamstring muscle strain injuries.
Sebastian Kaufmann, Olaf Hoos, Aaron Beck, Fabian Fueller, Richard Latzel, and Ralph Beneke
Purpose: To evaluate the metabolic relevance of type of locomotion in anaerobic testing by analyzing and comparing the metabolic profile of the Bosco Continuous Jumping Test (CJ30) with the corresponding profile of the Wingate Anaerobic Test (WAnT). Methods: A total of 11 well-trained, male team-sport athletes (age = 23.7 [2.2] y, height = 184.1 [2.8] cm, weight = 82.4 [6.4] kg) completed a CJ30 and WAnT each. During the WAnT, power data and revolutions per minute were recorded, and during the CJ30, jump height and jumping frequency were recorded. In addition, oxygen uptake and blood lactate concentration were assessed, and metabolic profiles were determined via the PCr-LA-O2 method. Results: In the CJ30, metabolic energy was lower (109.3 [18.0] vs 143.0 [13.1] kJ, P < .001, d = −2.302), while peak power (24.8 [4.4] vs 11.8 [0.5] W·kg−1, P < .001, d = 3.59) and mean power (20.8 [3.6] vs 9.1 [0.5] W·kg−1, P < .001, d = 4.14) were higher than in the WAnT. The metabolic profiles of the CJ30 (aerobic energy = 20.00% [4.7%], anaerobic alactic energy [W PCr] = 45.6% [4.5%], anaerobic lactic energy = 34.4% [5.2%]) and the WAnT (aerobic energy = 16.0% [3.0%], anaerobic alactic W PCr = 34.5% [5.0%], anaerobic lactic energy = 49.5% [3.3%]) are highly anaerobic. Absolute energy contribution for the CJ30 and WAnT was equal in W PCr (49.9 [11.1] vs 50.2 [11.2] kJ), but anaerobic lactic energy (37.7 [7.7] vs 69.9 [5.3] kJ) and aerobic energy (20.6 [5.7] vs 23.0 [4.0] kJ) were higher in the WAnT. Mechanical efficiency was substantially higher in the CJ30 (37.9% [4.5%] vs 15.6% [1.0%], P < .001, d = 6.86), while the fatigue index was lower (18.5% [3.8%] vs 23.2% [3.1%], P < .001, d = −1.38) than in the WAnT. Conclusions: Although the anaerobic share in both tests is similar and predominant, the CJ30 primarily taxes the W PCr system, while the WAnT more strongly relies on the glycolytic pathway. Thus, the 2 tests should not be used interchangeably, and the type of locomotion seems crucial when choosing an anaerobic test for a specific sport.
Laura Hottenrott, Sascha Ketelhut, Christoph Schneider, Thimo Wiewelhove, and Alexander Ferrauti
Postexercise recovery is a fundamental component for continuous performance enhancement. Due to physiological and morphological changes in aging and alterations in performance capacity, athletes of different ages may recover at different rates from physical exercise. Differences in body composition, physiological function, and exercise performance between men and women may also have a direct influence on restoration processes. Purpose: This brief review examines current research to indicate possible differences in recovery processes between male and female athletes of different age groups. The paper focuses on postexercise recovery following sprint and endurance tests and tries to identify determinants that modulate possible differences in recovery between male and female subjects of different age groups. Results: The literature analysis indicates age- and sex-dependent differences in short- and long-term recovery. Short-term recovery differs among children, adults, and masters. Children have shorter lactate half-life and a faster cardiac and respiratory recovery compared to adults. Additionally, children and masters require shorter recovery periods during interval bouts than trained adults. Trained women show a slower cardiac and respiratory recovery compared to trained men. Long-term recovery is strongly determined by the extent of muscle damage. Trained adults tend to have more extensive muscle damage compared to masters and children. Conclusion: The influence of age and sex on the recovery process varies among the different functional systems and depends on the time of the recovery processes. Irrespective of age and sex, the performance capacity of the individual determines the recovery process after high-intensity and endurance exercise.
Jonpaul Nevin and Paul Smith
Purpose: The aim of the following case study was to evaluate the effectiveness of a 30-week concurrent strength and endurance training program designed to prepare a trained H4 male handcyclist (aged 28 y, bilateral, above knee amputee, and body mass 65.6 kg) for a 1407-km ultra-endurance handcycling challenge. Methods: This observational case study tracked selected physiological measures, training intensity distribution, and total training load over the course of a 30-week concurrent training protocol. Furthermore, the athlete’s performance profile during the ultra-endurance challenge was monitored with power output, cadence, speed, and heart rate recorded throughout. Results: Findings revealed considerable improvements in power output at a fixed blood lactate concentration of 4 mmol·L−1 (+25.7%), peak aerobic power output (+18.9%), power-to-mass ratio (+18.3%), relative peak oxygen uptake (+13.9%), gross mechanical efficiency (+4.6%), bench press 1-repetition maximum (+4.3%), and prone bench pull 1-repetition maximum (+14.9%). The athlete completed the 1407-km route in a new handcycling world record time of 89:55 hours. Average speed was 18.7 (2.1) km·h−1; cadence averaged 70.0 (2.6) rpm, while average power output was 67 (12) W. In terms of internal load, the athlete’s average heart rate was 111 (11) beats per minute. Conclusion: These findings demonstrate how a long-term concurrent strength and endurance training program can be used to optimize handcycling performance capabilities in preparation for an ultra-endurance cycling event. Knowledge emerging from this case study provides valuable information that can guide best practices with respect to handcycling training for ultra-endurance events.
Myles C. Dennis, Paul S.R. Goods, Martyn J. Binnie, Olivier Girard, Karen E. Wallman, Brian T. Dawson, and Peter Peeling
Purpose: This study aimed to assess the influence of graded air temperatures during repeated-sprint training in hypoxia (RSH) on performance and physiological responses. Methods: Ten well-trained athletes completed one familiarization and 4 experimental sessions at a simulated altitude of 3000 m (0.144 FIO2) above sea level. Air temperatures utilized across the 4 experimental sessions were 20°C, 25°C, 30°C, and 35°C (all 50% relative humidity). The participants performed 3 sets of 5 × 10 seconds “all-out” cycle sprints, with 20 seconds of active recovery between sprints and 5 minutes of active recovery between sets (recovery intensity = 120 W). Core temperature, skin temperature, pulse oxygen saturation, heart rate, rating of perceived exertion, and thermal sensation were collected. Results: There were no differences between conditions for peak power, mean power, and total work in each set (P > .05). There were no condition × time interaction effects for any variables tested. The peak core temperature was highest at 30°C (38.06°C [0.31°C]). Overall, the pulse oxygen saturation was higher at 35°C than at 20°C (P < .001; d < 0.8), 25°C (P < .001; d = 1.12 ± 0.54, large), and 30°C (P < .001; d = 0.84 ± 0.53, large). Conclusion: Manipulating air temperature between 20°C and 35°C had no effect on performance or core temperature during a typical RSH session. However, the pulse oxygen saturation was preserved at 35°C, which may not be a desirable outcome for RSH interventions. The application of increased levels of ambient heat may require a different approach if augmenting the RSH stimulus is the desired outcome.
Valentin Bottollier, Matt R. Cross, Nicolas Coulmy, Loïc Le Quellec, and Jacques Prioux
Purpose: The purpose of this study was to determine the test–retest reliability of the 80s-slide-test in well-trained alpine ski racers. Methods: The sample consisted of 8 well-trained alpine ski racers (age = 17.8 [0.7] y old; height = 1.80 [0.09] m; body mass = 72.1 [9.5] kg) who performed a lab-based maximal graded test on cycle ergometer and three 80s-slide-tests in 4 separate sessions. The 80s-slide-test consisting of maximal push-offs performed for 80s on a 8-ft slide board. Oxygen uptake (
Ed Maunder, Deborah K. Dulson, and David M. Shaw
Purpose: Considerable interindividual heterogeneity has been observed in endurance performance responses following induction of a ketogenic diet (KD). It is plausible that a physiological stress response in the period following the dramatic dietary shift associated with transition to a KD may explain this heterogeneity. Methods: In a randomized, crossover study design, 8 trained male runners completed an incremental exercise test and ran to exhaustion at 70%VO2max before and after a 31-day rigorously controlled habitual diet or KD intervention, and recorded heart rate variability (root mean square of the sum of successive differences in R–R intervals [rMSSD]) upon waking each morning along with the recovery–stress questionnaire for athletes each week. Data were analyzed using linear mixed models. Results: A significant reduction in rMSSD was observed in the KD (−9.77 [4.03] ms, P = .02), along with an increase in day-to-day variability in rMSSD (2.1% [1.0%], P = .03). The reduction in rMSSD in the KD for the subgroup of individuals exhibiting impaired exercise capacity following induction of the KD approached significance (Δ −22  ms, P = .06, N = 4); whereas no effect was observed in those who exhibited unchanged exercise capacity (Δ 5  ms, P = .61, N = 4). No main effects were observed for recovery–stress questionnaire for athletes. Conclusions: Our data suggest those working with endurance athletes transitioning onto a KD may consider using noninvasive, inexpensive resting heart rate variability measures to gain individual-level insights into the likely short-term effects on exercise capacity.
Pedro L. Valenzuela, Fernando Rivas, and Guillermo Sánchez-Martínez
Purpose: To describe the effects of COVID-19 lockdown and a subsequent retraining on the training workloads, autonomic responses, and performance of a group of elite athletes. Methods: The training workloads and heart rate variability (assessed through the log-transformed root mean square of successive R–R intervals) of 7 elite badminton players were registered daily during 4 weeks of normal training (baseline), 7 to 10 weeks of lockdown, and 6 to 8 weeks of retraining. Physical performance was assessed at baseline and after each phase by means of a countermovement jump and the estimated squat 1-repetition maximum. Results: A reduction in training workloads was observed in all participants during the lockdown (−63.7%), which was accompanied by a reduced heart rate variability in all but one participant (−2.0%). A significant reduction was also observed for countermovement jump (−6.5%) and 1-repetition maximum performance (−11.5%), which decreased in all but one participant after the lockdown. However, after the retraining phase, all measures returned to similar values to those found at baseline. At the individual level, there were divergent responses, as exemplified by one athlete who attenuated the reduction in training workloads and increased her performance during the lockdown and another one who markedly reduced his workload and performance, and got injured during the retraining phase. Conclusions: Although there seems to be a large interindividual variability, COVID-19 lockdown is likely to impose negative consequences on elite athletes, but these detrimental effects might be avoided by attenuating reductions in training workloads and seem to be overall recovered after 6 to 8 weeks of retraining.