Fall risk is high for older adults with dementia. Gait impairment contributes to increased fall risk, and gait changes are common in people with dementia, although the reliable assessment of gait is challenging in this population. This study aimed to develop an automated approach to performing gait assessments based on gait data that is collected frequently and unobtrusively, and analysed using computer vision methods. Recent developments in computer vision have led to the availability of open source human pose estimation algorithms, which automatically estimate the joint locations of a person in an image. In this study, a preexisting pose estimation model was applied to 1066 walking videos collected of 31 older adults with dementia as they walked naturally in a corridor on a specialized dementia unit over a two week period. Using the tracked pose information, gait features were extracted from video recordings of gait bouts and their association with clinical mobility assessment scores and future falls data was examined. A significant association was found between extracted gait features and a clinical mobility assessment and the number of future falls, providing concurrent and predictive validation of this approach.

Background Gait impairments contribute to falls in people with dementia. In this study, we used a vision-based system to record episodes of walking over a 2-week period as participants moved naturally around their environment, and from these calculated spatiotemporal, stability, symmetry, and acceleration gait features. The aim of this study was to determine whether features of gait extracted from a vision-based system are associated with falls, and which of these features are most strongly associated with falling. Methods Fifty-two people with dementia admitted to a specialized dementia unit participated in this study. Thirty different features describing baseline gait were extracted from Kinect recordings of natural gait over a 2-week period. Baseline clinical and demographic measures were collected, and falls were tracked throughout the participants’ admission. Results A total of 1,744 gait episodes were recorded (mean 33.5 ± 23.0 per participant) over a 2-week baseline period. There were a total of 78 falls during the study period (range 0–10). In single variable analyses, the estimated lateral margin of stability, step width, and step time variability were significantly associated with the number of falls during admission. In a multivariate model controlling for clinical and demographic variables, the estimated lateral margin of stability (p = .01) was remained associated with number of falls. Conclusions Information about gait can be extracted from vision-based recordings of natural walking. In particular, the lateral margin of stability, a measure of lateral gait stability, is an important marker of short-term falls risk.