Automatic Fusion of Inertial Sensors and Clinical Risk Factors for Accurate Fall Risk Assessment during Balance Assessment

Abstract

Falls are a complex, common and costly problem in the older adult population. Deficits in balance and postural control have long been associated with falls risk. Assessment of balance in a clinical setting can require expensive, non-portable equipment as well as specialist expertise. This study aims to evaluate the accuracy of a combination of (1) self-reported clinical falls risk factors and (2) IMU based measurement of standing balance for assessment of falls risk in community dwelling older adults. 277 participants (99 male, 178 female) received a Comprehensive Geriatric Assessment (CGA) and standing balance tests (eyes open and eight eyes closed) while wearing a lumbar-mounted IMU. Results obtained through classifier fusion, validated using nested cross-validation, suggest that IMU data combined with clinical risk factors are significantly more accurate (67.9%) than clinical risk data alone (58.1%) or the Berg balance scale (BBS) (59.2%). We report a method that may be suitable for assessment of falls risk and standing balance in the home environment.