Journal Logo

E-40 Free Communication/Poster - Research Methodology Friday, June 3, 2016, 7: 30 AM - 12: 30 PM Room: Exhibit Hall A/B

Evidence of Convergent Validity for Measuring Free-Living Walking Using Wearable Devices

2773 Board #296 June 3, 9

30 AM - 11

00 AM

Lee, Miyoung; Choi, Ji-Yeob; Lee, Kwanghee; Min, Jiye; Kim, Jaemyung; Choi, Mun Cheong

Author Information
Medicine & Science in Sports & Exercise: May 2016 - Volume 48 - Issue 5S - p 780
doi: 10.1249/01.mss.0000487342.55249.86
  • Free

PURPOSE: To determine the convergent validity of consumer-based wearable devices for assessing walking during free-living activities.

METHODS: A total of 48 healthy adults (10 males, 50.3±10.6 yrs & BMI 24.9±2.5; 38 females, 53.3±7.4 yrs & BMI 24.4±2.9) participated. The consumer-based wearable devices included Actigraph GT3X+(AT), BodyMedia Armband (BM), Fitbit Charge (FC), Fitbit One (FO), Jawbone up (JB), Misfit (MF), and the Nike Fuelband (NF) worn on wrist, waist, or upper arm. The Omron HJ720IT pedometer served as the reference measure. While simultaneously wearing all devices, walking time (6.64±.80min), steps (13.28±1.22), and speed (1.43±.20m/sec) were assessed (10 m walk). Participants also completed selected free-living activities including moving boxes, cleaning desks, mopping, walking up & down stairs, and outdoor walking on hills. On average these activities took 40min to complete. Descriptive statistics, correlation coefficients (r) and paired t-test among the wearable devices and the Omron pedometer were calculated using SPSS 20.

RESULTS: It was unable to examine the accuracy of the JB due to constant synchronizing errors. Walking Time, Steps, and Speed showed moderate correlations with the Omron (r = .39, .55, & .43, respectively); but low correlations with wearable devices r = .11 to .23(Time), -.10 to .37(Steps), and -.08 to -.23(Speed). Step counts measured by the devices during free-living were as follows: FC=178±238, NF=1595±156, MF=1628±283, BM=1667±191, FO=1735±174, and AT=1778±180. The Omron (reference) recorded 1749±169 steps. The correlation coefficients between the step counts from the wearable devices and the Omron during free-living were AT=.91, FO=.91, NF =.61, FC=.59, BM=.44, and MF=-.15. All wearable devices showed statistically significant differences when compared to the Omron (t = -2.2 to -14.3, p <0.05).

CONCLUSIONS: : Of the devices examined, AT and FO were the most valid for measuring walking steps during free-living activities. On the other hand, quality control and measurement error associated with JB and MF suggest they should not be used.

This study were supported by National Research Foundation of Korea (NRF-2014R1A1A3049992)

© 2016 American College of Sports Medicine