Predicting Falls in Older Adults with an Accelerometer-Based Smartphone App

Abstract

Abstract Purpose This study investigated the accuracy of an accelerometer-based smartphone application for predicting the risk of falls in older adults. Methods Eighty community-dwelling older adults (70.05 ± 4.5 years) were assessed using an accelerometer-based smartphone application for balance assessment at baseline, which included 1) the Modified Clinical Test of Sensory Interaction in Balance (MCTSIB), 2) a single-leg stance (SLST), and 3) limit of stability (LOS) test. The fall incidence during a 6-month follow-up was recorded. The area under the receiver operating characteristic curve (AUC) was used to determine accuracy. Results The accuracy in predicting falls of all assessments was high (AUC = 0.78–0.99). The MCTSIB had the highest AUC (0.99) compared with the SLST (0.86) and the LOS test (0.78). The MCTSIB had a cutoff score of 8.04, sensitivity of 100%, and specificity of 98%. Conclusion The accelerometer-based smartphone application could predict falls in older adults with excellent accuracy.