Validation of pattern-recognition monitors in children using doubly labeled water

Area of Science:

• Pediatric obesity research
• Wearable sensor technology
• Human energy metabolism

Background:

• Accurate assessment of physical activity and energy expenditure (EE) is crucial for childhood obesity prevention research.
• Wearable monitors offer a promising approach for objective measurement in free-living conditions.
• Evaluating the validity of these devices in youth populations is essential.

Purpose of the Study:

• To validate the accuracy of two SenseWear Armband (SWA) monitors (Pro3 and Mini) in assessing physical activity and energy expenditure in children.
• To compare the performance of two different algorithms (version 2.2 and the newly developed 5.0) on these monitors.
• To assess validity under free-living conditions in a youth population.

Main Methods:

• Twenty-eight healthy children (10-16 years) wore both SWA monitors for 14 days.
• Total EE estimates were calculated using SenseWear software algorithms version 2.2 and 5.0.
• Monitor-derived EE was compared against doubly labeled water (DLW) methodology using ANOVA, correlation, and Bland-Altman plots.

Main Results:

• The SenseWear Armband 5.0 algorithm demonstrated significantly lower mean absolute percentage errors (10.9-11.7%) compared to the 2.2 algorithm (18.3-20.7%).
• High correlations (>0.90) were observed across all comparisons, but Bland-Altman plots indicated overestimation bias at higher EE levels.
• While the 5.0 algorithm improved group comparisons, individual variability in EE estimation persisted.

Conclusions:

• The SenseWear Armband 5.0 algorithms provide more accurate group-level estimates of energy expenditure in children than version 2.2.
• Despite improved algorithms, significant individual variability in energy expenditure measurements requires further investigation.
• These findings contribute to refining wearable technology for pediatric obesity research.