Why detection of cycling is important in activity monitoring

Introduction: Walking and cycling are effective means to increase activity-related energy expenditure (EE), the modifiable component of total EE. Activity monitors fixed to the trunk have not been able to distinguish these activities. However, cycling is expected to coincide with lower movement intensity compared to walking with comparable EE [1]. This study aimed to evaluate the effects of misclassification of cycling with an activity monitor. Methods: Physical activity of a convenience sample of 17 adults was measured for 37 measurement days with the DynaPort MoveMonitor (MM). The MM detects lying, sitting, standing and locomotion and the movement intensity (MI) per activity class. In addition we used a new method to detect cycling with an accuracy of 88.8%. We analysed to which activity classes cycling periods were assigned. Results: MM software classified cycling as walking (84%), sitting (5%), standing (3%) or shuffling (7%). With the new algorithm, 21037 walking and 218 cycling periods were identified. Mean duration of these periods was 9.8 s for walking and 355 s for cycling. Duration of cycling was 31.4% of the total locomotion duration. Weighted mean MI was 0.21 g for walking and 0.15 g for cycling, which is 28% lower than for walking. Conclusions: Cycling duration was almost 1/3 of total locomotion and cycling periods had a much longer duration in Dutch adults. Given the lower MI during cycling and possibly higher levels of actual EE in cycling compared to walking, EE will be underestimated when cycling is classified as walking in the existing MM software. [1] Brandes et al. Estimating energy expenditure from raw accelerometry in three types of locomotion. Med Sci Sports Exerc. 2012 Nov; 44(11):2235-42.