Research Study Abstract

Accelerometer Derived Movement Pattern Differences Across Age

  • Presented on 2011

Introduction Pattern recognition classification algorithms have successfully identified energy cost and activity type from hip-worn accelerometers (ACC) [1,2]. These algorithms evaluate attributes of the acceleration signal during high frequency sampling. ACC output similarities and differences across activity types are well characterized [3]. An often overlooked area of potential difference lies in movement pattern disparities across age. The purpose of this study was to examine movement pattern differences across age during treadmill (TM) walking and activities of daily living (ADL).

Methods Participants (N=118) underwent measures of stature, mass, and percent body fat (%BF) and were classified into 3-groups (Gp1, 18-39yrs, n=48; Gp2, 40-59yrs, n=39; Gp3, 60-79yrs, n=31). Each participant completed 7-TM walking activities (26.8-107.2 m/min in 13.4 m/min increments), and 5-ADLs (1: computer, 2: sweeping/mopping, 3: vacuuming, 4: moving boxes, and 5: walking/intermittent stair climbing) for 4-7 min each. During activities a tri-axial ACC (Actigraph GT3x) was worn on the left ankle (LA), left hip (LH), and left wrist (LW). Each ACC was initialized to record in 1-s epochs, with mean counts derived from averaging the last 2-min of TM and 4-min of ADL activity. ANOVA and GLM were used for unadjusted and adjusted (stature and %BF) age group comparisons on movement patterns.

Results All groups significantly differed by age (p<0.001), height (p<0.05), and %BF (p<0.001). The magnitude of mean acceleration significantly increased in unadjusted and adjusted analyses (p<0.001) across TM walking speeds for the LA, LH and LW. During these activities there were no significant differences in ACC output within each TM walking speed across all age group categories. The magnitude of mean unadjusted and adjusted acceleration significantly differed across each ADL within each age group category (p<0.05). For ADLs 2, 3, 4 and 5, ACT output was significantly lower (p<0.05) in age Gp3 compared with Gps1 and 2, in the magnitude of 5-13% lower for LA, 3-11% lower for LH, and 7-24% lower for LW.

Discussion and Conclusion ACC derived movement patterns during paced activities did not differ across age group classifications. During self-paced ADLs, ACC movement was decreased in the old, with differences becoming more pronounced with higher energy cost activities. Findings suggest that with older age movement patterns during ADLs become unique from other ages, where individual’s likely self pace as a function of individual physiological attributes. Future work is warranted to evaluate predictive capabilities of advanced modeling approaches as a function of age and physiological function.

References [1] Staudenmayer, J, Pober, D, Crouter, S, Bassett, D, & Freedson, P. An artificial neural network to estimate physical activity energy expenditure and identify physical activity type from an accelerometer. J. App. Physiol. 2009; 107:1300-1307. [2] Rothney, M, Neumann, M, Beziat, A, & Chen, K. An artificial neural network model of energy expenditure using non-integrated acceleration signals. J. App. Physiol. 2007; 103:1419-1427. [3] Basset, D, Ainsworth, B, Swartz, A, Strath S, O’Brien W, & King G. Validity of four motion sensors in measuring moderate intensity physical activity. Med Sci. Sports Exerc. 2000; 32: S471-S480.


  • Strath, S. J.
  • Keenan, K. G.
  • Hart, T. L.
  • Grimm, E. K.
  • Miller, N. E.
  • Swartz, A. M.


  • Physical Activity & Health Research Laboratory, University of Wisconsin-Milwaukee, Milwaukee, USA

Presented at

ICAMPAM- Glasgow 2011