Comparison of physical activity measurement using smartphone-based and wrist worn accelerometers

Purpose: The objective of this study was to compare accelerometer data taken with an onboard accelerometer in a smartphone (Droid A855) with accelerations recorded by a stand-alone accelerometer (ActiGraph GT3x+).

Methods: A smartphone application was developed to record the time varying acceleration from the Droid’s onboard three dimensional accelerometer. Twenty-seven adolescents wore an ActiGraph on their non-dominant wrist. Each adolescent played three smartphone-based games (e.g., Fish Out of Water, Space Rayders, and Color Hunt) that were designed to promote physical activity (PA). Each game was played 30 minutes a day on each, 4 days in a single week. Activity counts and the sum of the absolute value of the difference (SAD) calculations were employed to calculate PA from the Droid data. Regression analyses were used to assess the correlation between the Droid data and that collected by the ActiGraph.

Results: Droid-based assessment of physical activity in terms of activity counts (R2=0.63) and SAD (R2=0.57) were significantly related (p<.05). While the activity counts and SAD values increased with Actigraph counts for all game, the strongest relationships were with ActiGraph activity measures for Fish Out of Water, a game that encouraged fast walking/running in short distances (R2=0.64 activity counts; R2=0.68 SAD).

Conclusions: This study demonstrated that a smartphone accelerometer application can be used to record PA data that is correlated with the ActiGraph. These results suggest that the onboard accelerometers in ubiquitous computing devices like mobile phones may be a reliable method to assess PA levels and contribute to activity promotion and monitoring.