Share this article on:

Accelerometer Prediction of Energy Expenditure: Vector Magnitude Versus Vertical Axis

HOWE, CHERYL A.1; STAUDENMAYER, JOHN W.2; FREEDSON, PATTY S.1

Medicine & Science in Sports & Exercise: December 2009 - Volume 41 - Issue 12 - pp 2199-2206
doi: 10.1249/MSS.0b013e3181aa3a0e
Applied Sciences

ABSTRACT: It is suggested that triaxial accelerometers (RT3) are superior to single-plane accelerometers for predicting energy expenditure (EE).

Purpose: To compare the RT3 uniaxial and triaxial prediction of activity EE (AEE) during treadmill activities (TM) and activities of daily living (ADL).

Methods: Two hundred and twelve subjects (aged 20-60 yr) completed TM speeds of 1.34, 1.56, and 2.23 m·s−1 at 0% and 3% grades, stair ascent/descent, moving a box, and two randomly assigned ADL. Subjects wore a portable indirect calorimeter to measure EE to calculate AEE by subtracting resting metabolic rate. Acceleration counts in the vertical (V), medial-lateral, and anterior-posterior planes were collected in a single RT3 secured to the hip. Predicted AEE (RT3AEE) was estimated from vector magnitude (VM) counts using a proprietary algorithm. A paired t-test compared RT3AEE versus AEE. The relationship among V and VM counts and AEE was examined using linear regression analyses.

Results: RT3 overestimated AEE for all activities combined, overestimated for TM (9.0%), and underestimated for ADL (34.3%; P < 0.001). The R2 values between RT3AEE and AEE for TM and ADL were R2 = 0.78 and R2 = 0.15, respectively. The RT3 underestimated activity with greater upper body movements by 24.4%-64.5% (P < 0.001). V and VM counts were similarly related to AEE (R2 = 0.35) and RT3AEE (R2 = 0.83-0.89).

Conclusions: Although the RT3 did not accurately predict AEE from accelerometer counts, stronger relationships existed between predicted and measured AEE for TM compared with ADL. Compared with V counts, using VM counts to predict AEE did not significantly improve the relationship between counts and AEE. Analytic techniques beyond linear regression with VM as a covariate or with counts from each axis entering the model separately may improve estimates of AEE from triaxial accelerometers.

1Department of Kinesiology, University of Massachusetts, Amherst, MA; and 2Department of Math and Statistics, University of Massachusetts, Amherst, MA

Address for correspondence: Patty S. Freedson, Ph.D., Department of Kinesiology, University of Massachusetts, 30 Eastman Lane, 162 Totman Bldg, Amherst, MA 01003; E-mail: psf@kin.umass.edu.

Submitted for publication January 2009.

Accepted for publication April 2009.

©2009The American College of Sports Medicine