Abstract
BACKGROUND: PortionSize is a smartphone application that utilizes augmented reality and portion size templates to facilitate users' food intake estimates. MyFitnessPal is a commercial application that tracks food intake based on user's portion size estimates. Both applications provide real-time feedback on dietary intake, with PortionSize also providing feedback on intake of MyPlate food groups. OBJECTIVES: This study aimed to test validity of both applications for accurately measuring food intake compared with weighed food. METHODS: In a within-subjects randomized counterbalanced design, we tested the validity of both applications for accurately measuring food intake compared with weighed food. In semicontrolled free-living settings, adults received a cooler each day that contained preweighed foods and were instructed to consume only food and beverages provided. Participants used PortionSize and MyFitnessPal during 3-d periods separated by ∼1-wk with prior training on each application. After study periods, participants completed a User Preference Survey (UPS). Two one-sided t-tests, with ±18% equivalence bounds, were used to test whether energy intake estimated with applications was statistically equivalent to weighed intake. RESULTS: The analysis sample (n = 44; 1 dropped) consisted of 55% females, with a mean (standard deviation) age of 44 y (11) y. PortionSize (P = 0.032) and MyFitnessPal (P < 0.001) energy intake (kcal) estimates were equivalent to weighed intake (mean percent error 8.0% compared with 3.7%, respectively). MyFitnessPal had a smaller relative absolute error (P < 0.001) for energy intake compared with PortionSize. PortionSize and weighed food group estimations were equivalent for protein intake (P = 0.002). Although estimates for vegetables, fruits, and dairy were not equivalent, the magnitude of error was relatively low (<0.5 cup Eq). For the UPS, 82% of participants preferred using MyFitnessPal compared with PortionSize (18%; P < 0.001). CONCLUSIONS: In semicontrolled free-living settings, both applications provided equivalent energy intake estimates compared with weighed foods; however, PortionSize showed greater error for energy and lower user preference than MyFitnessPal, highlighting the need for improvements to decrease error and enhance user experience. This trial was registered at clinicaltrials.gov as NCT05166226.