An investigation of clinical and sensor-based fall-risk assessment in community-dwelling older adults
thesisposted on 2022-10-17, 08:11 authored by Valerie Power
Accurate, efficient methods of assessing fall-risk are required to identify at-risk community-dwelling older adults and implement timely falls prevention interventions. Sensor-based fall-risk assessment (SBFRA) methods have been developed to objectively assess and quantify fall-risk by analysing functional task performance, but research exploring their clinical applications is lacking. The current research aimed to investigate if SBFRA could perform clinically-meaningful fall-risk assessment in community-dwelling older adults (i.e. could it accurately classify older adults according to their level of fall-risk), and to explore its use among high-risk older adults participating in a community-based falls prevention intervention. Following thorough examination of current evidence and issues of feasibility, clinical and SBFRA was carried out among High-Risk (n=38) and Low-Risk (n=33) groups of older adults in the community, the High-Risk group being participants in a community-based falls prevention intervention. An array of sensor-derived variables extracted from static and dynamic task performances distinguished between High-Risk and Low-Risk groups; among them, novel sensor-derived variables that quantified standing balance and TUG performance strategy and quality, as well as simple sensor-derived gait variables e.g. cadence, mean step time. Some improvements in sensor-based indicators of standing balance performance were observed following intervention. Simple sensor-derived gait variables classified High-Risk participants more accurately than clinical tools e.g. cadence (sensitivity/specificity: 90.9%) and mean step time (sensitivity: 87.9%, specificity: 90.9%) versus Berg Balance Scale (sensitivity: 86.8%, specificity: 81.8%). Classification tree models comprised of 1) clinical, 2) sensor-based variables and 3) both combined, exhibited excellent fall-risk classification properties, with 95.8% accuracy for all models. This research confirms that SBFRA can be used to perform clinically-meaningful fall-risk assessment among community-dwelling older adults. With further research to develop specific evidence-based and user-friendly methods, SBFRA could be used to augment clinical fall-risk assessments, thereby assisting healthcare providers with clinical reasoning and outcome measurement in community falls prevention.
First supervisorClifford, Amanda M.
Second supervisorvan de Ven, Pepijn
Third supervisorNelson, John
Department or School
- Allied Health