Classification of muscle activity based on effort level during constant pace running

Abstract

During running, psychologic and physiologic changes are manifested in the perception of effort, muscle properties and movement strategies. The latter two aspects are expressed as changes in electromyographic (EMG) activity. This paper tests the hypothesis that the EMG signals change in a systematic way during a run and that these changes are related to the effort level of the runner. Fifteen female recreational runners performed 1-h treadmill runs at a constant speed (95% of speed at ventilatory threshold). EMG signals were recorded from four muscles (tibialis anterior, gastrocnemius medialis, vastus lateralis, and semitendinosus). The wavelet transformed EMG data were used to discriminate between different effort phases of running using a support vector machine (SVM) approach. The effect of the penalty parameter, C, and cross validation folds, n, used were evaluated and found to have little influence on the outcome. Recognition rates of >80% were achieved for all C and n values across all muscles. Average recognition rates were: TA - 89.2, GM - 88.3%, VL - 84.6% and ST - 94.0%. These results suggest that selected lower extremity EMG signals using wavelet-based methods contained highly systematic differences that could be used by the SVM to discriminate between the low- and high-effort stages of prolonged running.