Skip to Main Content
A methodology for mapping in-home gait speed (IGS), measured unobtrusively and continuously in the homes of older adults, to Timed-Up-and-Go (TUG) time is presented. A Kinect-based gait system was used to collect in-home gait data on 15 older adults over time periods of up to 16 months. Concurrently, the participants completed a monthly clinician administered fall risk assessment protocol that included TUG and habitual gait speed (HGS) tests. A theoretical analysis of expected performance is presented, and the performance of the IGS-based TUG estimates is compared against that of estimates based on HGS measured at the same time as the TUG. Results indicate that the IGS-based estimates are as accurate as the HGS-based estimates as compared to the observed TUG times. After filtering the TUG times to reduce noise, the IGS-based estimates are more accurate. The mapping of in-home sensor data to well studied domains facilitates clinical interpretation of the in-home data.