A spasticity assessment method for voluntary movement using data fusion and machine learning

The assessment of spasticity under voluntary movement is helpful for the therapist to comprehensively assess patient's dyskinesia. However, current researches focus on evaluation based passive motion. We propose a new method evaluating active Our following three steps: (i) Empirical Mode Decomposition (EMD) used reduce involuntary noise in patients' movement; (ii) Extract segments each muscle feature extract and fusion; (iii) Use machine learning methods evaluate degree spasm patients. To investigates feasibility proposed this paper, An experiment elbow flexion extension against gravity designed, electromyographic signal brachioradialis (BR), biceps brachialis (BB), triceps (TB) motion data 13 subjects were collected. compared classification effect filter method, window length classifier type. Moreover, we analyze improvement by fusion. results showed that random forest with 256 ms had best (F1-score = 0.952). Compared 0.756) or only 0.7053), presented paper better accuracy. Result demonstrated our method. This study can assist doctors spasmodic state movement, has application potential wearable devices.