A Machine Learning Technique to Analyze Surface EMG Signals in Normal and Diabetic Subjects

Abstract

The diabetes mellitus results in many musculoskeletal complications and detection of these problems is important in treatment, prevention of disability and improving the quality of life. Many attempts have been made to effectively acquire and analyze the surface electromyogram (sEMG) signals to understand the musculoskeletal problems in diabetes. The development of methodologies to extract the effective features from sEMG still remains a primary challenge. Previous studies have demonstrated that the sEMG signals have nonlinear characteristics. In this study, by combining the nonlinear time-series analysis and the time-frequency domain methods, we propose the wavelet based method to extract the effective features of sEMG signals. The sEMG signals were first acquired and analyzed by the wavelet transform and the features were obtained. Then, these features were used as the input vectors to artificial neural network (ANN) classifier to discriminate diabetic or non-diabetic subject. The results show that significance of proposed feature extraction method with classification of diabetic and nondiabetic subjects with an accuracy of 97.06%.This indicates that the proposed approach can provide important insight into the nonlinear characteristics and the time-frequency domain features of sEMG signals and is suitable for classifying the normal subjects from diabetic subjects. By comparing with other existing methods, the proposed method exhibited more robustness and higher classification accuracy.