A stepwise science-industry collaboration to optimize the calculation of energy expenditure during walking and running with a consumer-based activity device

Physical activity (PA) and sedentary behavior (SB) have important implications for health benefits. A growing number of people use consumer available wearables such as activity trackers, which claim to objectively monitor PA and SB in free-living conditions. These devices could provide essential information to understand the influence of behavior on health. This understanding assumes that available consumer products correctly monitor PA in the everyday life. A general approach in science is to validate such activity devices in a controlled environment. The classical procedure to investigate criterion validity is to examine new devices based on the gold standard. To our knowledge, the resulting validation data are not often analyzed and shared with manufacturers to further develop and improve the activity device. The current study can be seen as a validation study to check the criterion validity of a consumer-level activity device. The novelty of this study was the application of a stepwise approach to optimize the calculations of a consumer available activity device (i.e., Activ8; www.activ8all.com/product/activ8-professional-activity-monitor/) for estimating energy expenditure (EE) in walking and running. Forty adults (27 males and 13 females) participated in three substudies. Each substudy consisted of several walking and running activities in which EE was simultaneously measured with indirect calorimetry (as reference value) and the Activ8 activity device. EE values at each walking and running speed were compared to identify the accuracy of the Activ8 device. After completion of the first and second substudies, the results were shared and discussed with the manufacturer of Activ8. Next, the calculations for EE were adapted to the indirect calorimetry values to improve accuracy. In the second and third substudies, the modifications were tested, and results were used to further optimize the calculation of EE. The results of this study show an improved correlation between EE measured by indirect calorimetry and the Activ8 activity device (R2 from 0.91 to 0.95); a decrease in differences between substudy A and substudy B considering EE measured (indirect calorimetry) and calculated (Activ8 calculation) was observed. The second modification in the calculation showed a further increase in correlation (R2 from 0.95 to 0.97) between the measured and calculated EE; however, the absolute difference between the two values increased. The results from a validation study are valuable to use for further adaptation of accelerometer device calculations. A stepwise science-industry collaboration can improve the calculation accuracy and may be a practical approach for validation studies in which human movement scientists and technology manufacturers work together to successfully improve the validity and accuracy of consumer-based activity devices.