Mark Pfeiffer

Applications of neural networks in training science.

Training science views itself as an integrated and applied science, developing practical measures founded on scientific method. Therefore, it demands consideration of a wide spectrum of approaches and methods. Especially in the field of competitive sports, research questions are usually located in complex environments, so that mainly field studies are drawn upon to obtain broad external validity. Here, the interrelations between different variables or variable sets are mostly of a nonlinear character. In these cases, methods like neural networks, e.g., the pattern recognizing methods of Self-Organizing Kohonen Feature Maps or similar instruments to identify interactions might be successfully applied to analyze data. Following on from a classification of data analysis methods in training-science research, the aim of the contribution is to give examples of varied sports in which network approaches can be effectually used in training science. First, two examples are given in which neural networks are employed for pattern recognition. While one investigation deals with the detection of sporting talent in swimming, the other is located in game sports research, identifying tactical patterns in team handball. The third and last example shows how an artificial neural network can be used to predict competitive performance in swimming.

A Markov Chain Model of Elite Table Tennis Competition

The evaluation of the structure of sports performance is one of the important functions of diagnostics in competitive sport. Especially in game sports, it is important to obtain diagnostic information on competition because of the interactive process between the two teams or players. This interaction cannot be simulated or replicated in training or test situations. When it comes to table tennis, performance diagnostics offers many different techniques and methods to analyze a game. In this context, problems mostly occur in adequately modelling the game. Moreover, the approaches lack a scope of uniform performance criteria. According to a stochastic performance diagnostic concept (Markov Chain) for game sports, we developed four different state-transition-models to describe tactical behaviour in table tennis: (1) game action, (2) stroke position, (3) stroke direction and (4) stroke technique. Afterwards we formalized these models by means of finite Markov Chain (stochastic modelling) to determine the relevance of tactical behaviour to performance by simulation. In this study, 152 games of the top 50 ranked male players in the world were analyzed. The results suggest that in international-level table tennis, the diagonal game from backhand to backhand (“stroke position”), strokes into the long backhand zone (“stroke direction”), and especially the topspin stroke for shake-hand players in general (“stroke technique”), are the most important winning tactical strategies. In comparison with traditional methods used in elite table tennis the approach described here is valuable in quantifying and comparing the relevance of various tactical behaviours to performance.

Validity of the acute recovery and stress scale: training monitoring of the German junior national field hockey team

The aim of the present study was to examine the sensitivity of the Acute Recovery and Stress Scale (ARSS). This new psychometric questionnaire was developed to assess the physical, mental, emotional, and overall recovery and stress states of athletes. During a five-day field hockey training camp of the German Junior National Field Hockey Team (n = 25) the ARSS was administered every morning and evening. The study indicated swift reactions of the scores of the physical and general factors as well as stability of scores for the emotional factors in accordance with the training schedule. The straining effect of the camp was best reflected by the adaptations of the scales Physical Performance Capability(F2.9, 60.3 = 10.0, p< 0.001) and Muscular Stress(F4, 84 = 16.7, p < 0.001). The results support the ability of the ARSS to monitor recovery-stress (im-)balances in this sample. Thus, the questionnaire has shown to be a sensitive and practical tool that might be suitable for elite sport settings.

Markers for Routine Assessment of Fatigue and Recovery in Male and Female Team Sport Athletes during High-Intensity Interval Training

Aim Our study aimed to investigate changes of different markers for routine assessment of fatigue and recovery in response to high-intensity interval training (HIIT). Methods 22 well-trained male and female team sport athletes (age, 23.0 ± 2.7 years; V̇O2max, 57.6 ± 8.6 mL·min·kg−1) participated in a six-day running-based HIIT-microcycle with a total of eleven HIIT sessions. Repeated sprint ability (RSA; criterion measure of fatigue and recovery), countermovement jump (CMJ) height, jump efficiency in a multiple rebound jump test (MRJ), 20-m sprint performance, muscle contractile properties, serum concentrations of creatinkinase (CK), c-reactive protein (CRP) and urea as well as perceived muscle soreness (DOMS) were measured pre and post the training program as well as after 72 h of recovery. Results Following the microcycle significant changes (p < 0.05) in RSA as well as in CMJ and MRJ performance could be observed, showing a decline (%Δ ± 90% confidence limits, ES = effect size; RSA: -3.8 ± 1.0, ES = -1.51; CMJ: 8.4 ± 2.9, ES = -1.35; MRJ: 17.4 ± 4.5, ES = -1.60) and a return to baseline level (RSA: 2.8 ± 2.6, ES = 0.53; CMJ: 4.1 ± 2.9, ES = 0.68; MRJ: 6.5 ± 4.5, ES = 0.63) after 72 h of recovery. Athletes also demonstrated significant changes (p < 0.05) in muscle contractile properties, CK, and DOMS following the training program and after the recovery period. In contrast, CRP and urea remained unchanged throughout the study. Further analysis revealed that the accuracy of markers for assessment of fatigue and recovery in comparison to RSA derived from a contingency table was insufficient. Multiple regression analysis also showed no correlations between changes in RSA and any of the markers. Conclusions Mean changes in measures of neuromuscular function, CK and DOMS are related to HIIT induced fatigue and subsequent recovery. However, low accuracy of a single or combined use of these markers requires the verification of their applicability on an individual basis.

Muscle mechanical properties of strength and endurance athletes and changes after one week of intensive training

The study investigates whether tensiomyography (TMG) is sensitive to differentiate between strength and endurance athletes, and to monitor fatigue after either one week of intensive strength (ST) or endurance (END) training. Fourteen strength (24.1±2.0years) and eleven endurance athletes (25.5±4.8years) performed an intensive training period of 6days of ST or END, respectively. ST and END groups completed specific performance tests as well as TMG measurements of maximal radial deformation of the muscle belly (Dm), deformation time between 10% and 90% Dm (Tc), rate of deformation development until 10% Dm (V10) and 90% Dm (V90) before (baseline), after training period (post1), and after 72h of recovery (post2). Specific performance of both groups decreased from baseline to post1 (P<0.05) and returned to baseline values at post2 (P<0.05). The ST group showed higher countermovement jump (P<0.05) and shorter Tc (P<0.05) at baseline. After training, Dm, V10, and V90 were reduced in the ST (P<0.05) while TMG changes were less pronounced in the END. TMG could be a useful tool to differentiate between strength and endurance athletes, and to monitor fatigue and recovery especially in strength training.

Assessment of neuromuscular function after different strength training protocols using tensiomyography

Abstract De Paula Simola, RÁ, Harms, N, Raeder, C, Kellmann, M, Meyer, T, Pfeiffer, M, and Ferrauti, A. Assessment of neuromuscular function after different strength training protocols using tensiomyography. J Strength Cond Res 29(5): 1339–1348, 2015—The purpose of the study was to analyze tensiomyography (TMG) sensitivity to changes in muscle force and neuromuscular function of the muscle rectus femoris (RF) using TMG muscle properties after 5 different lower-limb strength training protocols (multiple sets; DS = drop sets; eccentric overload; FW = flywheel; PL = plyometrics). After baseline measurements, 14 male strength trained athletes completed 1 squat training protocol per week over a 5-week period in a randomized controlled order. Maximal voluntary isometric contraction (MVIC), TMG measurements of maximal radial displacement of the muscle belly (Dm), contraction time between 10 and 90% of Dm (Tc), and mean muscle contraction velocities from the beginning until 10% (V10) and 90% of Dm (V90) were analyzed up to 0.5 (post-train), 24 (post-24), and 48 hours (post-48) after the training interventions. Significant analysis of variance main effects for measurement points were found for all TMG contractile properties and MVIC (p < 0.01). Dm and V10 post-train values were significantly lower after protocols DS and FW compared with protocol PL (p = 0.032 and 0.012, respectively). Dm, V10, and V90 decrements correlated significantly to the decreases in MVIC (r = 0.64–0.67, p ⩽ 0.05). Some TMG muscle properties are sensitive to changes in muscle force, and different lower-limb strength training protocols lead to changes in neuromuscular function of RF. In addition, those protocols involving high and eccentric load and a high total time under tension may induce higher changes in TMG muscle properties.

Acute effects of psychological relaxation techniques between two physical tasks

ABSTRACT The concept of recovery strategies includes various ways to achieve a state of well-being, prevent underrecovery syndromes from occurring and re-establish pre-performance states. A systematic application of individualised relaxation techniques is one of those. Following a counterbalanced cross-over design, 27 sport science students (age 25.22 ± 1.08 years; sports participation 8.08 ± 3.92 h/week) were randomly assigned to series of progressive muscle relaxation, systematic breathing, power nap, yoga, and a control condition. Once a week, over the course of five weeks, their repeated sprint ability was tested. Tests (6 sprints of 4 s each with 20 s breaks between them) were executed on a non-motorised treadmill twice during that day intermitted by 25 min breaks. RM-ANOVA revealed significant interaction effects between the relaxation conditions and the two sprint sessions with regard to average maximum speed over all six sprints, F(4,96) = 4.06, P = 0.004, = 0.15. Post-hoc tests indicated that after systematic breathing interventions, F(1,24) = 5.02, P = 0.033, = 0.18, participants performed significantly better compared to control sessions. As the focus of this study lied on basic mechanisms of relaxation techniques in sports, this randomised controlled trial provides us with distinct knowledge on their effects, i.e., systematic breathing led to better performances, and therefore, seems to be a suited relaxation method during high-intensity training.

Assessment of Fatigue and Recovery in Male and Female Athletes After 6 Days of Intensified Strength Training.

Abstract Raeder, C, Wiewelhove, T, Simola, RÁDP, Kellmann, M, Meyer, T, Pfeiffer, M, and Ferrauti, A. Assessment of fatigue and recovery in male and female athletes after 6 days of intensified strength training. J Strength Cond Res 30(12): 3412–3427, 2016—This study aimed to analyze changes of neuromuscular, physiological, and perceptual markers for routine assessment of fatigue and recovery in high-resistance strength training. Fourteen male and 9 female athletes participated in a 6-day intensified strength training microcycle (STM) designed to purposefully overreach. Maximal dynamic strength (estimated 1 repetition maximum [1RMest]; criterion measure of fatigue and recovery); maximal voluntary isometric strength (MVIC); countermovement jump (CMJ) height; multiple rebound jump (MRJ) height; jump efficiency (reactive strength index, RSI); muscle contractile properties using tensiomyography including muscle displacement (Dm), delay time (Td), contraction time (Tc), and contraction velocity (V90); serum concentration of creatine kinase (CK); perceived muscle soreness (delayed-onset muscle soreness, DOMS) and perceived recovery (physical performance capability, PPC); and stress (MS) were measured before and after the STM and after 3 days of recovery. After completing the STM, there were significant (p ⩽ 0.05) performance decreases in 1RMest (%[INCREMENT] ± 90% confidence limits, ES = effect size; −7.5 ± 3.5, ES = −0.21), MVIC (−8.2 ± 4.9, ES = −0.24), CMJ (−6.4 ± 2.1, ES = −0.34), MRJ (−10.5 ± 3.3, ES = −0.66), and RSI (−11.2 ± 3.8, ES = −0.73), as well as significantly reduced muscle contractile properties (Dm, −14.5 ± 5.3, ES = −0.60; V90, −15.5 ± 4.9, ES = −0.62). After days of recovery, a significant return to baseline values could be observed in 1RMest (4.3 ± 2.8, ES = 0.12), CMJ (5.2 ± 2.2, ES = 0.28), and MRJ (4.9 ± 3.8, ES = 0.32), whereas RSI (−7.9 ± 4.5, ES = −0.50), Dm (−14.7 ± 4.8, ES = −0.61), and V90 (−15.3 ± 4.7, ES = −0.66) remained significantly reduced. The STM also induced significant changes of large practical relevance in CK, DOMS, PPC, and MS before to after training and after the recovery period. The markers Td and Tc remained unaffected throughout the STM. Moreover, the accuracy of selected markers for assessment of fatigue and recovery in relation to 1RMest derived from a contingency table was inadequate. Correlational analyses also revealed no significant relationships between changes in 1RMest and all analyzed markers. In conclusion, mean changes of performance markers and CK, DOMS, PPC, and MS may be attributed to STM-induced fatigue and subsequent recovery. However, given the insufficient accuracy of markers for differentiation between fatigue and recovery, their potential applicability needs to be confirmed at the individual level.

Can the Lamberts and Lambert Submaximal Cycle Test Indicate Fatigue and Recovery in Trained Cyclists

The Lamberts and Lambert Submaximal Cycle Test (LSCT) is a novel test designed to monitor performance and fatigue/recovery in cyclists. Studies have shown the ability to predict performance; however, there is a lack of studies concerning monitoring of fatigue/recovery. In this study, 23 trained male cyclists (age 29 ± 8 y, VO2max 59.4 ± 7.4 mL · min(-1) · kg(-1)) completed a training camp. The LSCT was conducted on days 1, 8, and 11. After day 1, an intensive 6-day training period was performed. Between days 8 and 11, a recovery period was realized. The LSCT consists of 3 stages with fixed heart rates of 6 min at 60% and 80% and 3 min at 90% of maximum heart rate. During the stages, power output and rating of perceived exertion (RPE) were determined. Heart-rate recovery was measured after stage 3. Power output almost certainly (standardized mean difference: 1.0) and RPE very likely (1.7) increased from day 1 to day 8 at stage 2. Power output likely (0.4) and RPE almost certainly (2.6) increased at stage 3. From day 8 to day 11, power output possibly (-0.4) and RPE likely (-1.5) decreased at stage 2 and possibly (-0.1) and almost certainly (-1.9) at stage 3. Heart-rate recovery was likely (0.7) accelerated from day 1 to day 8. Changes from day 8 to day 11 were unclear (-0.1). The LSCT can be used for monitoring fatigue and recovery, since parameters were responsive to a fatiguing training and a following recovery period. However, consideration of multiple LSCT variables is required to interpret the results correctly.

A New Method to Individualize Monitoring of Muscle Recovery in Athletes

PURPOSE Assessment of muscle recovery is essential for the daily fine-tuning of training load in competitive sports, but individual differences may limit the diagnostic accuracy of group-based reference ranges. This article reports an attempt to develop individualized reference ranges using a Bayesian approach comparable to that developed for the Athlete Biological Passport. METHODS Urea and creatine kinase (CK) were selected as indicators of muscle recovery. For each parameter, prior distributions and repeated-measures SDs were characterized based on data of 883 squad athletes (1758 data points, 1-8 per athlete, years 2013-2015). Equations for the individualization procedure were adapted from previous material to allow for discrimination of 2 physiological states (recovered vs nonrecovered). Evaluation of classificatory performance was carried out using data from 5 consecutive weekly microcycles in 14 elite junior swimmers and triathletes. Blood samples were collected every Monday (recovered) and Friday according to the repetitive weekly training schedule over 5 wk. On the group level, changes in muscle recovery could be confirmed by significant differences in urea and CK and validated questionnaires. Group-based reference ranges were derived from that same data set to avoid overestimating the potential benefit of individualization. RESULTS For CK, error rates were significantly lower with individualized classification (P vs group-based: test-pass error rate P = .008; test-fail error rate P < .001). For urea, numerical improvements in error rates failed to reach significance. CONCLUSIONS Individualized reference ranges seem to be a promising tool to improve accuracy of monitoring muscle recovery. Investigating application to a larger panel of indicators is warranted.