One-year persistence of individual gait patterns identified in a follow-up study - A call for individualised diagnose and therapy.

Although a hunch about the individuality of human movements generally exists, differences in gait patterns between individuals are often neglected. To date, only a few studies distinguished individual gait patterns in terms of uniqueness and emphasised the relevance of individualised diagnoses and therapy. However, small sample sizes have been a limitation on identifying subjects based on gait patterns, and little is known about the permanence of subject-specific characteristics over time. The purpose of this study was (1) to prove the uniqueness of individual gait patterns within a larger sample and (2) to prove the long-term permanence of individual gait patterns. A sample of 128 healthy participants each walked a distance of 10m barefoot 10 times. Two force plates recorded the ground reaction forces during a double step at a self-selected walking speed. A subsample of 46 participants repeated this procedure after a period of 7-16 months. The application of support vector machines resulted in classification rates of 99.8% (1278 out of 1280) and 99.4% (914 out of 920) for the initial subject-classification and the subsample follow-up-classification, respectively. The results showed that gait patterns based on time-continuous ground reaction forces were unique to an individual and could be differentiated from those of other individuals. Support vector machines classified gait patterns to the corresponding individual almost error-free. Hence, human gait is not only different between individuals but also exhibits unique individual characteristics that are persistent over years. Our findings provide evidence for the individual nature of human walking and emphasise the need to evaluate individualised clinical approaches for diagnoses and therapy.

Daily changes of individual gait patterns identified by means of support vector machines.

Despite the common knowledge about the individual character of human gait patterns and about their non-repeatability, little is known about their stability, their interactions and their changes over time. Variations of gait patterns are typically described as random deviations around a stable mean curve derived from groups, which appear due to noise or experimental insufficiencies. The purpose of this study is to examine the nature of intrinsic inter-session variability in more detail by proving separable characteristics of gait patterns between individuals as well as within individuals in repeated measurement sessions. Eight healthy subjects performed 15 gait trials at a self-selected speed on eight days within two weeks. For each trial, the time-continuous ground reaction forces and lower body kinematics were quantified. A total of 960 gait patterns were analysed by means of support vector machines and the coefficient of multiple correlation. The results emphasise the remarkable amount of individual characteristics in human gait. Support vector machines results showed an error-free assignment of gait patterns to the corresponding individual. Thus, differences in gait patterns between individuals seem to be persistent over two weeks. Within the range of individual gait patterns, day specific characteristics could be distinguished by classification rates of 97.3% and 59.5% for the eight-day classification of lower body joint angles and ground reaction forces, respectively. Hence, gait patterns can be assumed not to be constant over time and rather exhibit discernible daily changes within previously stated good repeatability. Advantages for more individual and situational diagnoses or therapy are identified.

Time scales of adaptive behavior and motor learning in the presence of stochastic perturbations.

In this paper, the major assumptions of influential approaches to the structure of variability in practice conditions are discussed from the perspective of a generalized evolving attractor landscape model of motor learning. The efficacy of the practice condition effects is considered in relation to the theoretical influence of stochastic perturbations in models of gradient descent learning of multiple dimension landscapes. A model for motor learning is presented combining simulated annealing and stochastic resonance phenomena against the background of different time scales for adaptation and learning processes. The practical consequences of the model's assumptions for the structure of practice conditions are discussed, together with their implications for teaching and coaching.

A quantitative dynamical systems approach to differential learning: self-organization principle and order parameter equations.

Differential learning is a learning concept that assists subjects to find individual optimal performance patterns for given complex motor skills. To this end, training is provided in terms of noisy training sessions that feature a large variety of between-exercises differences. In several previous experimental studies it has been shown that performance improvement due to differential learning is higher than due to traditional learning and performance improvement due to differential learning occurs even during post-training periods. In this study we develop a quantitative dynamical systems approach to differential learning. Accordingly, differential learning is regarded as a self-organized process that results in the emergence of subject- and context-dependent attractors. These attractors emerge due to noise-induced bifurcations involving order parameters in terms of learning rates. In contrast, traditional learning is regarded as an externally driven process that results in the emergence of environmentally specified attractors. Performance improvement during post-training periods is explained as an hysteresis effect. An order parameter equation for differential learning involving a fourth-order polynomial potential is discussed explicitly. New predictions concerning the relationship between traditional and differential learning are derived.

A pattern recognition approach for the quantification of horse and rider interactions.

REASONS FOR PERFORMING STUDY: Interactions of various systems were investigated in several studies of dynamic systems, but the interactions between horse and rider have not yet been documented. These interactions include the rider's ability to control the horse, adapt to the horse and maintain both participants' body position. An optimum interaction is also adapted to the individual nature of the horse. OBJECTIVE: To identify rider-horse interactions by means of artificial neural nets analysing the time-continuous pattern. METHODS: Fourteen horses were measured trotting on hand, and ridden at working trot with a professional and a recreational rider using a 3D high speed video system (120 Hz)1. Angles were calculated after low pass filtering (5-20 Hz). Horse movements were described by 2D angles, angular velocities, and angular accelerations of variables of the right body side: hind and front fetlock, head, back and the summation angle of carpus, elbow, and shoulder, the summation angle of hock, stifle, and hip. Distances between the trajectories of the feature vectors in an N = 11 x 11 Kohonen map were determined and analysed by means of a cluster analysis. RESULTS: Depending on the variables included, both rider specific as well as horse specific movement patterns could be identified. The time courses of the head angle indicate a movement pattern mainly dominated by the rider, whereas the time courses of variables of the hind fetlock and hock in most cases did not show differences between the conditions with, and without, rider. The skill of the professional rider could be documented with a higher adaptation to the horse's movement pattern. CONCLUSION AND POTENTIAL RELEVANCE: The presented time course oriented approach provides a sensitive tool in order to quantify the interaction of rider and horse.

Applications of artificial neural nets in clinical biomechanics.

The purpose of this article is to provide an overview of current applications of artificial neural networks in the area of clinical biomechanics. The body of literature on artificial neural networks grew intractably vast during the last 15 years. Conventional statistical models may present certain limitations that can be overcome by neural networks. Artificial neural networks in general are introduced, some limitations, and some proven benefits are discussed.

Identification of individual walking patterns using time discrete and time continuous data sets.

Scientific studies typically treat data by studying effects of groups. Clinical therapy typically treats patients on a subject specific basis. Consequently, scientific and clinical attempts to help patients are often not coordinated. The purposes of this study were (a) to identify subject and group specific locomotion characteristics quantitatively, using time discrete and time continuous data and (b) to assess the advantages and disadvantages of the two approaches. Kinematic and kinetic gait pattern of 13 female subjects walking in dress shoes with different heel heights (14, 37, 54 and 85 mm) were analysed. The results of this study showed that subject specific gait characteristics could be better identified with the time continuous than with the time discrete approach. Thus, the time continuous approach using artificial networks is an effective tool for identifying subject and group specific locomotion characteristics.