Biomecânica da corrida em diferentes faixas etárias: classificação baseada na técnica árvore de decisão e análise de codificação vetorial modificada

Abstract

Running is a popular exercise and the elderly population practicing this sport has increased. However, this activity is associated with a risk of lower limb injuries and female runners are twice as likely to develop these injuries. The majority of studies that compared age groups used classical statistics. The decision tree approach is used to classify patterns in data sets and has shown to be better than the support vector machine (SVM) on other sorts of data. Dependent variables have tended to focus on discrete data from isolated joints. Coordination variability (CV) quantifies the variety of segment movement patterns and is linked to a healthy motor system. The aims of this thesis were to evaluate the capacity of the decision tree to discriminate runners of different age groups, to compare the performance of the decision tree to the SVM and to compare CV among male and female runners, separately, during running. Kinematic and electromyography analysis were assessed. The decision tree was used to discriminate the age groups and a modified vector coding technique was used to investigate segment CV. The results show that the decision tree approach was capable of discriminating the different age groups. Also, for the male runners, the younger runners presented a higher CV than the middle-aged and the older. And for the female runners, the younger runners presented a lower CV than the middle-aged runners and a similar coordination variability when compared to the older group. The study findings indicate that decision tree approach had a better performance in discriminating the age groups than the SVM. Also, aging influences dynamic function for male runners and, lastly, female runners appear to maintain their CV during running regardless of age.