Automatic classification of gait patterns using a smart rollator and the BOSS model

Abstract

Nowadays, the risk of falling in older adults is a major concern due to the severe consequences it brings to socio-economic and public health systems. Some pathologies cause mobility problems in the aged population, leading them to fall and, thus, reduce their autonomy. Other implications of ageing involve having different gait patterns and walking speed. In this paper, a non-invasive framework is proposed to study gait in elder people using data collected by a smart rollator, the i-Walker. The analysis presented in this article uses a feature extraction method and a spectral embedding to represent the information and Bayesian clustering for the knowledge discovery. The algorithm considers raw data from the i-Walker sensors along with the calculated walking speed of each individual, which has been already used in clinical studies to assess physical and cognitive status of older adults. The results obtained demonstrate that the proposed analysis has the potential to separate in clusters the people of the two groups of interest: young people and geriatric.