Consumer fitness trackers useful for monitoring changes in daily rhythm, physical activity

While wrist-worn consumer physical activity trackers (PATs) tend to overestimate step counts, they are also sensitive to upper-limb movements and can potentially be used to track changes over time in the wearers’ physical activity, according to a recent study.

Aside from measuring step counts, “[o]ur results also suggest that, unlike the self-reported method, the wearable PAT can be expected to measure the individual’s life rhythm from their sleep and wakefulness patterns without the need for conscious efforts for daily recording and sufficient cognitive ability,” the researchers said.

Forty participants (mean age 32.65 years, 20 women) were given two PAT devices: one worn on the wrist and another around the waist. Participants were then subjected to two experiments. The first assessed the effects of upper-limb activity on device-measured step count, while the second evaluated the accuracy of wristband-type PATs at measuring sleep times.

Results from the first experiment showed that while performing upper-limb tasks, the waist-worn PAT did not log step counts, as expected. However, the wrist-worn devices did record step counts, suggesting that in normal circumstances, wristband consumer devices would tend to overestimate step counts as they also take into consideration the movement of the upper limbs. [PLoS One 2022;doi:10.1371/journal.pone.0271155]

Of note, only the type of task performed was significantly correlated with step counts (p<0.001); meanwhile, neither the distance travelled nor the pace at which the upper limbed moved as it performed the tasks had no such effect.

In the second experiment, participants self-reported an average sleep time of 413.43 minutes, while PAT devices measured a mean of 414.85 minutes of sleep per night. Bland-Altman analysis revealed a mean error of –1.42 minutes, indicating that the wristband devices overestimated sleep by <2 minutes.

Regression analysis showed no evidence of systematic error. This was confirmed by Pearson’s correlation, which demonstrated a strong and significant association between self-reported and device-measured sleep duration (coefficient, 0.93; p<0.001).

“It is known that if appropriate support is not provided to the patient population undergoing rehabilitation, mortality and the degree of long-term care may worsen due to a decrease in physical activity and disrupted sleep cycle, leading to irregular daytime sleeping,” the researchers said, adding that using monitoring devices which need to be removed periodically to be charged could be detrimental to adherence.

“The wristband-type PAT offers a feasible solution to capture these situations more easily and accurately, which allows for early countermeasure and verification of individual state changes obtained by appropriate intervention,” they said.

The present findings indicated that for use in such scenarios, data from wrist-worn devices need to be analysed in consideration of the patients’ upper-limb activity. “Nevertheless, the device is expected to accurately capture the within-subject changes in the amount of physical activity and the daily rhythm,” the researchers said.