Demonstrating the utility of a person-worn camera to assess multiple behaviors in different populations: Beyond accelerometers and food recalls

Purpose: To provide objective measures of physical activity (PA) and eating behaviors (EB) using a single device. Person-worn cameras (PWC) offer the opportunity to classify and characterize multiple concurrent activities and their contexts, simultaneously; allowing for the objective classification of contextual and time-specific activities.

Methods: Two distinct populations, overweight females (OF n=197 person-days) and university employees commuting via bicycle (CB n=90 person-days), wore a PWC for a minimum of 8 hours a day for 3-5 days along with a hip worn accelerometer. PWC images were annotated for position(e.g. sedentary, or walking/running) and activities (e.g. eating, TV watching, etc) using standardized procedures.

Results: Results are presented to demonstrate the utility of PWC images to objectively identify behaviors in different populations.

• OF showed no difference in sitting time between weekdays and weekends, compared to a 10% difference in CB group.

• Time spent cycling was double on weekdays vs weekends in the CB group (8.2% vs 4.3%) while accelerometry did not detect differences in activity level.

• Both the OF and CB group performed more household activity on weekends vs weekdays (10% and 14% respectively).

• In the CB group, 13% of eating time was in front of a TV/screen on weekends

• OF evidenced stable car use on weekdays vs. weekends while the CB group varied by as much as 9%.

Conclusions: This study demonstrates how a PWC can assess multiple behaviors concurrently, which may be useful in better understanding the context and timing of specific behaviors.