Table 6 Summary of activities and behavioral features during free-living instances in which proposed method over and underestimated EE with respect to Kerr et al.’s algorithm7.
From: Developing and comparing a new BMI inclusive energy expenditure algorithm on wrist-worn wearables
Activity State | No. of Instances | No. of Unique Participants | Behavioral Features |
|---|---|---|---|
Underestimation | |||
Walking | 87 | 13 | Hands in pocket, gait, holding bags, holding phone steady |
Walking, shopping | 12 | 1 | Pushing cart, holding bags |
Sitting, public transit, phone | 62 | 4 | Holding phone steady |
Sitting | 33 | 4 | Swiveling, adjusting posture |
Sitting, fidgeting | 29 | 5 | Adjusting devices |
Sitting, phone | 16 | 9 | Talking on phone, hand stationary near ear |
Bathroom | 2 | 1 | Urinating (male), hands against wall |
Talking | 21 | 7 | Slight body movements |
Eating | 17 | 4 | Scrolling on phone |
Camera Blocked | 67 | 11 | Unable to identify activities |
Overestimation | |||
Sitting, phone | 98 | 13 | Scrolling, tapping on phone, gesticulating |
Walking | 23 | 6 | Opening doors, arms dangling |
Driving | 10 | 3 | Wheel turning, manipulation of dashboard |
Walking, shopping | 8 | 2 | Reaching for items, adding items to cart/bag |
Sitting, fidgeting | 78 | 15 | Dependent on type of activity being performed |
Cleaning | 13 | 4 | Object manipulation |
Phone, TV | 34 | 4 | Scrolling on phone, tapping on phone |
Eating | 9 | 3 | Sitting, standing |
Fidgeting | 21 | 8 | Smoking, drinking, device adjustment, sitting |
Camera Blocked | 20 | 6 | Unable to identify activities |
