Machine learning (ML) techniques such as (deep) artificial neural networks (DNN) are solving very successfully a plethora of tasks and provide new predictive models for complex physical, chemical, biological and social systems. However, in most cases this comes with the disadvantage of acting as a black box, rarely providing information about what made them arrive at a particular prediction. This black box aspect of ML techniques can be problematic especially in medical diagnoses, so far hampering a clinical acceptance. The present paper studies the uniqueness of individual gait patterns in clinical biomechanics using DNNs. By attributing portions of the model predictions back to the input variables (ground reaction forces and full-body joint angles), the Layer-Wise Relevance Propagation (LRP) technique reliably demonstrates which variables at what time windows of the gait cycle are most relevant for the characterisation of gait patterns from a certain individual. By measuring the time-resolved contribution of each input variable to the prediction of ML techniques such as DNNs, our method describes the first general framework that enables to understand and interpret non-linear ML methods in (biomechanical) gait analysis and thereby supplies a powerful tool for analysis, diagnosis and treatment of human gait.

Traditional learning approaches are typically based on a linear understanding of causality where the same cause leads to the same effect. In recent years there has been increasing interest in the complexity of nature and living phenomena, with significant insights provided by models of change that are based on a nonlinear understanding of causality, where small causes can lead to big effects and vice versa. In this vein, learning processes seem to be more successful for inducing behavioral change when teaching processes deviate from a linear approach. The differential learning approach takes advantage of fluctuations in a complex system by increasing them through 'no repetition' and 'constantly changing movement tasks' which add stochastic perturbations. Previous research has provided much evidence on the superiority of a differential learning approach for learning single movement techniques, in comparison to repetition-and correctionoriented approaches. In this pilot study, the parallel acquisition and learning of two movement techniques in the sport of football are the objective of investigation. One traditionally trained group and two differentially trained groups (blocked and random) trained for 4 weeks, twice a week, on ball control and shooting at goal tasks. Results supported previous work and revealed significant advantages for both differential groups in the acquisition phase as well as in the learning phase, compared to the traditional group. These data suggest that, instead of following a direct linear path towards the target of a 'to-be-learned' movement technique by means of numerous repetitions and corrections, a differential approach is more beneficial because it perturbs learners towards more functional movement patterns during practice.

Purpose: The aim of this study was to identify the effects of a differential-learning program, embedded in small-sided games, on the creative and tactical behavior of youth soccer players. Forty players from under-13 (U13) and under-15 (U15) were allocated into control and experimental groups and were tested using a randomized pretest to posttest design using small-sided games situations. Method: The experimental group participated in a 5-month differential-learning program embodied in small-sided games situations, while the control group participated in a typical small-sided games training program. In-game creativity was assessed through notational analyses of the creative components, and the players’ positional data were used to compute tactical-derived variables. Results: The findings suggested that differential learning facilitated the development of creative components, mainly concerning attempts (U13, small; U15, small), versatility (U13, moderate; U15, small), and originality (U13, unclear; U15, small) of players’ actions. Likewise, the differential-learning approach provided a decrease in fails during the game in both experimental groups (moderate). Moreover, differential learning seemed to favor regularity in pitch-positioning behavior for the distance between players’ dyads (U13, small; U15, small), the distance to the team target (U13, moderate; U15, small), and the distance to the opponent target (U13, moderate; U15, small). Conclusions: The differential-learning program stressed creative and positional behavior in both age groups with a distinct magnitude of effects, with the U13 players demonstrating higher improvements over the U15 players. Overall, these findings confirmed that the technical variability promoted by differential learning nurtures regularity of positioning behavior.