The use of smartphone technology to assess habitual dietary intake : A Literature review

Background: Collecting accurate dietary records that represents habitual intake is essential for investigating future individual risk of chronic diseases in clinical and epidemiological research. Due to the many limitations in conventional methods presented in existing reviews, the search for more accurate methods that can reduce respondent burden and costs has been encouraged. The use of smartphone technology to develop more reliable measures of dietary intake has been on the rise in response to increasing popularity of smartphones in the population. MyFitnessPal is a highly rated dietary application designed for weight loss and diet-tracking and is available commercially for free on iTunes and Google app stores. Aim: The present study aimed to validate MyFitnessPal to facilitate dietary assessment against a reference measure of 7-day estimated food diaries. Methods: A sample of 16 volunteers were randomly allocated to record their dietary intake using either the MyFitnessPal application or the 7-day estimated food diary before crossing-over to the alternative dietary assessment method for 7 consecutive days each. Mean daily intakes of nutrients recorded on both methods were compared using paired t-tests. Bland-Altman plots were used to assess for agreement between the two methods. The Goldberg and Black approach was used to identify implausible energy reporters by directly comparing energy intakes with energy expenditure. Results: Paired-t tests demonstrated no statistically significant difference between mean intakes of nutrients reported between the 7-day food diary and MyFitnessPal, 42 | P a g e except for water (p=0.004). However, all dietary variables were lower with the MFP method compared to the 7FD method. An analysis of the Bland-Altman plots presented agreement between the methods with small mean differences and minimal proportional bias. However, they showed wide limits of agreement signifying high levels of variability at the individual level. 75% of the sample were under-reporters with EIMFP:EE ratios and EI7FD:EE ratios below the 95% confidence limits. This study presented mean differences between EE and EI of -3.24 ± 2.20MJ for the 7FD and -3.75 ± 3.16MJ for the MFP app. Therefore, relative to EE, EI reported by the MFP was as accurate as the 7FD. Conclusion: At the group level, the MyFitnessPal app demonstrates potential but further investigation with a larger sample in addition to qualitative research of its acceptability is required to assess its feasibility as a dietary assessment method