TY - JOUR
T1 - A Multidimensional Approach to Develop Sway Index Using Gaussian Mixture Model
T2 - A Way of Postural Sway Measurement and Analysis in Different Age Groups
AU - Ghahramani, Maryam
AU - Ismail, Hafsa
AU - Goecke, Roland
N1 - Funding Information:
Manuscript received June 2, 2021; revised July 15, 2021; accepted August 10, 2021. Date of publication August 24, 2021; date of current version September 6, 2021. The work of Hafsa Ismail was supported by CSIRO/DATA61 for Ph.D. Scholarship. The Associate Editor coordinating the review process was Dr. Alessandra Galli. (Corresponding author: Maryam Ghahramani.) The authors are with the Faculty of Science and Technology, University of Canberra, Canberra, ACT 2617, Australia (e-mail: maryam.ghahramani@ canberra.edu.au). Digital Object Identifier 10.1109/TIM.2021.3107012
Publisher Copyright:
© 1963-2012 IEEE.
PY - 2021/9/5
Y1 - 2021/9/5
N2 - Many classical postural sway measures are based on linear analysis of different temporal characteristics of the body’s center of pressure (CoP). In some of the classical sway analysis methods, the anterior–posterior and the mediolateral aspects are analyzed separately. While these classical measures are found to be affected by aging, they cover different aspects of stability. Moreover, linear postural sway analysis is not efficient due to complexity of the human physiological functions. This study developed a single sway index based on the nonlinear analysis of the multidimensional CoP data and compared it in different participant age groups in different standing tests. The sway index performance was compared with six classical sway measures and six universal sway parameters. In all, 17 younger active lifestyle participants (31 ± 5.7), 12 participants age between 50 and 60 years (56.2 ± 3.3), and 32 older participants age 60 years and above (69.6 ± 6.2) were recruited for this study. Participants were asked to undergo three standing tests of double stance, single stance, and tandem stance all with eyes closed for 20 s. Using a global machine-learnt Gaussian mixture model, the multidimensional CoP data were clustered, and consequently an index was derived based on the results. Most classical and universal sway measures in the single stance and tandem stance were found to be significantly different in younger participants compared with the older ones. Our proposed sway index was significantly different in younger participants compared with 50–60 years participants in addition to older participants. The sway index also outperformed all classical and universal sway measures in the single and tandem stance tests with the sensitivity of 90.9% and 87.5%, the specificity of 82.4% and 84.3%, and AUC of 0.90 (95% CI, 0.81–1) and 0.91 (95% CI, 0.82–0.99), respectively. The findings demonstrated a strong potential of the sway index to be used as a single yet effective sway measurement.
AB - Many classical postural sway measures are based on linear analysis of different temporal characteristics of the body’s center of pressure (CoP). In some of the classical sway analysis methods, the anterior–posterior and the mediolateral aspects are analyzed separately. While these classical measures are found to be affected by aging, they cover different aspects of stability. Moreover, linear postural sway analysis is not efficient due to complexity of the human physiological functions. This study developed a single sway index based on the nonlinear analysis of the multidimensional CoP data and compared it in different participant age groups in different standing tests. The sway index performance was compared with six classical sway measures and six universal sway parameters. In all, 17 younger active lifestyle participants (31 ± 5.7), 12 participants age between 50 and 60 years (56.2 ± 3.3), and 32 older participants age 60 years and above (69.6 ± 6.2) were recruited for this study. Participants were asked to undergo three standing tests of double stance, single stance, and tandem stance all with eyes closed for 20 s. Using a global machine-learnt Gaussian mixture model, the multidimensional CoP data were clustered, and consequently an index was derived based on the results. Most classical and universal sway measures in the single stance and tandem stance were found to be significantly different in younger participants compared with the older ones. Our proposed sway index was significantly different in younger participants compared with 50–60 years participants in addition to older participants. The sway index also outperformed all classical and universal sway measures in the single and tandem stance tests with the sensitivity of 90.9% and 87.5%, the specificity of 82.4% and 84.3%, and AUC of 0.90 (95% CI, 0.81–1) and 0.91 (95% CI, 0.82–0.99), respectively. The findings demonstrated a strong potential of the sway index to be used as a single yet effective sway measurement.
KW - Postural sway
KW - Gaussian Mixture Model
KW - postural sway analysis
KW - fall risk assessment
KW - postural balance in older people
KW - Balance assessment
UR - https://ieeexplore.ieee.org/document/9521240
UR - http://www.scopus.com/inward/record.url?scp=85114508836&partnerID=8YFLogxK
U2 - 10.1109/TIM.2021.3107012
DO - 10.1109/TIM.2021.3107012
M3 - Article
SN - 0018-9456
VL - 70
SP - 1
EP - 11
JO - IEEE Transactions on Instrumentation and Measurement
JF - IEEE Transactions on Instrumentation and Measurement
M1 - 9521240
ER -