Objective: The focus of our cross-sectional study is on investigation of complex measures in terms of the fall risk (FR). The assessment was performed with data acquired from inertial sensors attached to the wrist during activities of daily living (ADL).

Background: Early assessment of the FR delivers clinically relevant information to the clinicians for the improvement of the fall prevention strategies and thus significantly reduces the costs in the healthcare. Nowadays, gait analysis based on complex measures has shown good performance in assessing one’s risk of falling, but there is still no evidence from the perspective of the wrist.

Methods: Data acquisition was performed with 239 adults (50-85 years, 52.9 % of female) with a wide range of physical performance (median habitual gait speed 1.12 m/s; range 0.38 m/s to 2.08 m/s). Subjects were split into two groups based on the previously validated FRAT-up score (cut-off point: 0.2). The sensor node equipped with the 3-axial inertial sensors was attached arbitrarily at dominant or non-dominant wrist for 7 consecutive days. A validated step detector was used for detection of steps and walking bouts. Due to smaller step rate variability, only walking bouts longer than 10 steps were submitted for the extraction of the following features: local dynamic stability (LDS), multi scale sample entropy (MSE) and harmonic ratio (HR). Mann-Whitney test was applied to each feature to analyse the difference between the corresponding means for the two defined groups, as well as for testing the feature’s side-dependence.

Results: The test has shown significantly higher (p = 0.02) LDS in the acceleration signal for fallers (0.38 ± 0.05) than for non-fallers (0.36 ± 0.03), while there was no difference between defined groups for the MSE and HR. Side-dependent analysis of the extracted features has shown significant difference for the LDS in acceleration and magnetic field in case when subjects were wearing the sensor node on their dominant wrist (p = 0.01 and p = 0.05, respectively). Results have shown no significant difference for the non-dominant wrist for any of the implemented features.

Conclusions: The proposed method for the assessment of the FR in ADL from the sensor node attached at the wrist shows statistically significant difference for the LDS feature extracted from the accelerometer and magnetometer signals independently on which side the sensor node was worn as well as only for the dominant wrist.

« Back to 2016 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/wrist-reveals-the-fall-risk-in-activities-of-daily-living/