Acceleration-based gait analysis for frailty assessment in older adults

•A fully automated system for gait analysis in the context of frailty.•Extraction of original features describing the older adults gait in regards to frailty status.•Assessment of frailty trajectory in elderly under unsupervised conditions during daily routine.•An added value in comparison with the standard frailty phenotype. Frailty in older individuals has been a hot research topic in the past two decades. This syndrome of physiological decline is characterized by damage to physical function. In the literature, walking speed and gait variability have been discussed and identified as major indicators of frailty. Short walking tests under supervised conditions are frequently considered when assessing the frailty status, where the subject is instructed to walk a certain distance, and the speed and temporospatial features are compared to thresholds. In this paper, we propose a more generalized and fully automated approach using wearable sensors, by suggesting a set of parameters extracted from acceleration signals. The interconnections between these variables, which are related to the gait quality and the frailty trajectory of the subject, are investigated. This study was done on older adults of diverse profiles, in free-living conditions during their daily routine, without any predefined protocol.