Wearable Technology to Quantify the Nutritional Intake of Adults: Validation Study

Abstract

Background: Wearable and mobile sensor technologies can be useful tools in precision nutrition research and practice, but few are reliable for obtaining accurate and precise measurements of diet and nutrition.

Objective: This study aimed to assess the ability of wearable technology to monitor the nutritional intake of adult participants. This paper describes the development of a reference method to validate the wristband's estimation of daily nutritional intake of 25 free-living study participants and to evaluate the accuracy (kcal/day) and practical utility of the technology.

Methods: Participants were asked to use a nutrition tracking wristband and an accompanying mobile app consistently for two 14-day test periods. A reference method was developed to validate the estimation of daily nutritional intake of participants by the wristband. The research team collaborated with a university dining facility to prepare and serve calibrated study meals and record the energy and macronutrient intake of each participant. A continuous glucose monitoring system was used to measure adherence with dietary reporting protocols, but these findings are not reported. Bland-Altman tests were used to compare the reference and test method outputs (kcal/day).

Results: A total of 304 input cases were collected of daily dietary intake of participants (kcal/day) measured by both reference and test methods. The Bland-Altman analysis had a mean bias of -105 kcal/day (SD 660), with 95% limits of agreement between -1400 and 1189. The regression equation of the plot was Y=-0.3401X+1963, which was significant (P<.001), indicating a tendency for the wristband to overestimate for lower calorie intake and underestimate for higher intake. Researchers observed transient signal loss from the sensor technology of the wristband to be a major source of error in computing dietary intake among participants.

Conclusions: This study documents high variability in the accuracy and utility of a wristband sensor to track nutritional intake, highlighting the need for reliable, effective measurement tools to facilitate accurate, precision-based technologies for personal dietary guidance and intervention.

Keywords: food intake; mobile health; mobile phone; validation study; wearable technology.

Figure 1

Healbe GoBe2 smartband.

Figure 2

Bland-Altman (mean difference) plot of estimated nutrient intakes (kilo/day) by the test and reference method (N=304). Solid lines represent upper-lower limits of agreement, and the dashed line represents bias.