Understanding Physical Activity Data Collection with Accelerometers
Author Information
Author(s): Paul David R, Kramer Matthew, Stote Kim S, Spears Karen E, Moshfegh Alanna J, Baer David J, Rumpler William V
Primary Institution: Beltsville Human Nutrition Research Center, Agricultural Research Service, United States Department of Agriculture
Hypothesis
Adherence estimates for activity monitors in adults will be strong, but missing data will negatively influence predictions of physical activity.
Conclusion
Imputation of missing activity monitor data can significantly improve estimates of physical activity.
Supporting Evidence
- The average adherence was estimated at 15.8 hours per day over approximately 11.7 days.
- Data loss during sleep resulted in biased estimates of physical activity.
- Imputation of missing data improved estimates of physical activity significantly.
Takeaway
This study shows that when people wear activity monitors, sometimes they take them off, which can make it hard to know how active they really are. But if we guess the missing data in a smart way, we can get a better idea of their activity levels.
Methodology
Participants wore activity monitors for 13-15 days, and data loss was simulated to analyze its effect on physical activity estimates.
Potential Biases
Potential bias due to non-compliance and variability in how subjects wore the monitors.
Limitations
The study may not generalize to populations outside of the Baltimore/Washington area, and the imputation methods may not be applicable for minute-by-minute data.
Participant Demographics
524 adults (262 men and 262 women) from the Baltimore, MD/Washington, DC area, with a range of ages and ethnicities.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.0001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website