Postural Balance for Selection of Martial Artists Using Machine Learning Techniques

Objectives: The purpose of this study was to classify participants, according to balance test scores, and to detect martial art athletes.Design: Measures of static and dynamic balance indices were obtained from ۴ tests.Setting: This research took place at a secondary school in Iran.Participants: Fifty healthy volunteers participated in this experiment.Main outcome measures: Due to differences in power and different pressures applied on joints and muscles, athletes in different sports and also non-athletes may have different grades in balance tests. There isn’t enough information on specific or non-specific balance in sports.Results: Balance test scores were used for inputs of classifiers where the applied methods included the support vector machine, k-nearest neighbors algorithm, and artificial neural network. Only by the result of ۴ tests, detection accuracy of ۹۰.۵% was achieved.Conclusion: Balance indices are good features for detection of martial art athletes. This may also be useful for talent identification in martial arts.

Keywords:

Artificial neural network , Balance , Classification , K-nearest neighbors , Support vector machine