Assessing the Frailty of Older People using Bluetooth Beacons Data

Abstract

Sensor-based quantitative assessment of frailty in older adults usually requires expensive wearables or ambient sensors. However, low cost activity tracking systems can be effective in discriminating frailty characteristics. In this work, a system for assessment of the frailty of older people is presented, based on the analysis of data describing the daily in-house activity. More specifically, room-to-room movements are recorded using a set of Bluetooth beacons, located in fixed positions inside each room. A smartphone is used to locate the timestamps of room transitions, and these are used to calculate the time spent in each room, thus generating the time intervals signal. Then, several features depicting the in-house mobility are extracted from each signal and are evaluated, in terms of correlation with the subject's frailty status. Furthermore, the features are used to train a classifier for frailty assessment. Two approaches are evaluated, with different number of levels of frailty, and the results indicate that this procedure can lead to reliable older people frailty assessments, especially when two levels of frailty are considered.