Abstract
A person's pattern of walking or gait is a defining characteristics of Parkinson's disease (PD). There is no cure for Parkinsonian gait, which continues to progress with time and varies among patients with PD. Detection of gait disorder can help identify PD at an early stage for optimal treatment and reducing adverse effects on the patient's well-being and daily activities. This paper presents the extraction of features of gait dynamics recorded from single wearable sensors, in frequency and spatial domains, for gait classification using sequential deep learning. Experimental results suggest that the proposed method is very promising for detecting gait in PD using single sensor-induced data in comparison with others using signals recorded from multiple wearable sensors. The implication is that the minimized deployment of sensors can avoid physical discomfort in patients and be cost-effective.