An alternative correlation based feature for measuring step symmetry for assessing fall risk in elderly persons
 
An alternative correlation based feature for measuring step symmetry for assessing fall risk in elderly persons 
 
Bart Jansen, Ivan Bautmans, Bart Van Keymolen, Rudi Deklerck, Tony Mets
 
Abstract 

Background: In tri-axial trunk accelerometry for assessing fall risk, the most common protocol consists of walking a fixed distance (e.g. 18 m.) on a straight line. Autocorrelation-based step-symmetry is an often used feature. Aim: To investigate differences in a new correlation-based feature in elderly according to fall risk. Methods: Forty elderly (>=70 years old) with increased fall risk (history of falls and/or Timed-Get-Up-and-Go-Test >15s and/or Tinetti-test Results: ICX0 is significantly different between the two groups (p=0.0017)in contrast to CC1 and CC2 (p=0.30 and p=0.77). However, |1- CC1| is significantly different (p=0.0015) as well as |1- CC2| (p=0.0018). Pearson's Correlation between CC1 and CC2 is low (R2=0.55). No significant relationship between ICX0 and CC1, CC2 or walking speed was found (respectively R2=0.017, R2=0.013 and R2=0.29). Conclusions: We propose ICX0 as a new outcome for assessing fall risk in elderly persons, which seems independent of walking speed and - although conceptually similar -not related to craniocaudal step symmetry.