Mobile and tablet apps, ubiquitous and pervasive computing, wearable computing and domotics for health.

Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author). Background: Nutrition-related apps are commonly used to provide information about the users’ dietary intake, but limited research has been performed on the evaluation of their reliability. Objective: To evaluate the relative reliability of popular nutrition-related apps for the assessment of energy and available macro- and micronutrients against a standard method. Methods: Dietary analysis of 24-hour weighed food records (n = 20) were compared between five nutrition-related apps: S Health, MyFitnessPal, FatSecret, Noom Coach and Lose It!, and DietPlan6 (reference method). Estimates of energy, macronutrients (carbohydrate, protein, fat, saturated fat and fibre) and micronutrients (sodium, calcium, iron, vitamin A and vitamin C) were compared using t-tests and Wilcoxon signed-rank tests, correlation coefficients and Bland-Altman plots. 24-hour weighed food records from 20 participants (female/male, 15/5; mean age, 36.3 years; mean body mass index, 22.9 kg/m2) recruited from previous controlled studies conducted at the Hugh Sinclair Unit of Human Nutrition, University of Reading, UK. Results: No significant difference in estimation of energy and saturated fat intake between DietPlan6 and the diet apps was observed. Estimates of protein and sodium intake were significantly lower using Lose It! and FatSecret than DietPlan6. Lose It! also gave significantly lower estimates for other reported outputs: carbohydrate, fat, fibre and sodium, compared with DietPlan6. For S Health and MyFitnessPal, calcium, iron and vitamin C were all significantly under-estimated compared with DietPlan6, although there was no significant difference for vitamin A. No other significant differences were observed between DietPlan6 and the apps. Correlation coefficients ranged from -0.12 for iron (S Health vs. DietPlan6) to 0.91 for protein (FatSecret vs DietPlan6). Noom Coach was limited to energy output but it had a high correlation with DietPlan6 (r=0.91) and S Health had the greatest variation of correlation, with energy at 0.79. Bland-Altman analysis revealed potential proportional bias for vitamin A. Conclusions: The findings suggest that the apps provide comparable estimates of energy and saturated fat intake compared with DietPlan6. With the exception of Lose It!, the apps also provided comparable estimates of carbohydrate, total fat and fibre. Two apps displayed a tendency to underestimate protein and sodium (FatSecret and Lose It!). Apart from vitamin A, the estimates of micronutrient intake (calcium, iron and vitamin C) by the two apps (S Health and MyFitnessPal) were inconsistent and less reliable. Lose It! was the less comparable app in relation to DietPlan6. As the use and availability of apps grows, this study helps clinicians and researchers to make better-informed decisions about using these apps in research and practice.

Background: Evidence suggests that eating behaviors and adherence to dietary guidelines can be improved using nutrition-related apps. Apps delivering personalized nutrition (PN) advice to users can provide individual support at scale with relatively low cost. Objective: This study aims to investigate the effectiveness of a mobile web app (eNutri) that delivers automated PN advice for improving diet quality, relative to general population food-based dietary guidelines. Methods: Nondiseased UK adults (aged >18 years) were randomized to PN advice or control advice (population-based healthy eating guidelines) in a 12-week controlled, parallel, single-blinded dietary intervention, which was delivered on the web. Dietary intake was assessed using the eNutri Food Frequency Questionnaire (FFQ). An 11-item US modified Alternative Healthy Eating Index (m-AHEI), which aligned with UK dietary and nutritional recommendations, was used to derive the automated PN advice. The primary outcome was a change in diet quality (m-AHEI) at 12 weeks. Participant surveys evaluated the PN report (week 12) and longer-term impact of the PN advice (mean 5.9, SD 0.65 months, after completion of the study). Results: Following the baseline FFQ, 210 participants completed at least 1 additional FFQ, and 23 outliers were excluded for unfeasible dietary intakes. The mean interval between FFQs was 10.8 weeks. A total of 96 participants were included in the PN group (mean age 43.5, SD 15.9 years; mean BMI 24.8, SD 4.4 kg/m2) and 91 in the control group (mean age 42.8, SD 14.0 years; mean BMI 24.2, SD 4.4 kg/m2). Compared with that in the control group, the overall m-AHEI score increased by 3.5 out of 100 (95% CI 1.19-5.78) in the PN group, which was equivalent to an increase of 6.1% (P=.003). Specifically, the m-AHEI components nuts and legumes and red and processed meat showed significant improvements in the PN group (P=.04). At follow-up, 64% (27/42) of PN participants agreed that, compared with baseline, they were still following some (any) of the advice received and 31% (13/42) were still motivated to improve their diet. Conclusions: These findings suggest that the eNutri app is an effective web-based tool for the automated delivery of PN advice. Furthermore, eNutri was demonstrated to improve short-term diet quality and increase engagement in healthy eating behaviors in UK adults, as compared with population-based healthy eating guidelines. This work represents an important landmark in the field of automatically delivered web-based personalized dietary interventions. Trial Registration: ClinicalTrials.gov NCT03250858; https://clinicaltrials.gov/ct2/show/NCT03250858

Background: The web-based EatWellQ8 food frequency questionnaire (FFQ) was developed as a dietary assessment tool for healthy adults in Kuwait. Validation against reliable instruments and assessment of its reproducibility are required to ensure the accuracy of the EatWellQ8 FFQ in computing nutrient intake. Objective: This study aims to assess the reproducibility and relative validity of the EatWellQ8 146-item FFQ, which included images of food portion sizes based on The Composition of Foods by McCance and Widdowson and food composition tables from Kuwait and the Kingdom of Bahrain, against a paper-based FFQ (PFFQ) and a 4-day weighed food record (WFR). Methods: Reproducibility of the EatWellQ8 FFQ was assessed using a test-retest methodology. Participants were required to complete the FFQ at 2 time points, 4 weeks apart. To assess the relative validity of the EatWellQ8 FFQ, a subset of the participants were asked to complete a PFFQ or a 4-day WFR 1 week after the administration of the EatWellQ8 FFQ. The level of agreement between nutrient and food group intakes was estimated by repeated EatWellQ8 FFQ administration. The EatWellQ8 FFQ, PFFQ, and 4-day WFR were also evaluated using the Bland-Altman methodology and classified into quartiles of daily intake. Crude unadjusted correlation coefficients were also calculated for nutrients and food groups. Results: A total of 99 Kuwaiti participants (64/99, 65% female and 35/99, 35% male) completed the study—53 participated in the reproducibility study and the 4-day WFR validity study (mean age 37.1 years, SD 9.9) and 46 participated in the PFFQ validity study (mean age 36.2 years, SD 8.3). Crude unadjusted correlations for repeated EatWellQ8 FFQs ranged from 0.37 to 0.93 (mean r=0.67, SD 0.14; 95% CI 0.11-0.95) for nutrients and food groups (P=.01). Mean cross-classification into exact agreement plus adjacent was 88% for nutrient intakes and 86% for food groups, and Bland-Altman plots showed good agreement for energy-adjusted macronutrient intakes. The association between the EatWellQ8 FFQ and PFFQ varied, with crude unadjusted correlations ranging from 0.42 to 0.73 (mean r=0.46, SD 0.12; 95% CI −0.02 to 0.84; P=.046). Mean cross-classification into exact agreement plus adjacent was 84% for nutrient intake and 74% for food groups. Bland-Altman plots showed moderate agreement for both energy and energy-controlled nutrient intakes. Crude unadjusted correlations for the EatWellQ8 FFQ and the 4-day WFR ranged from 0.40 to 0.88 (mean r=0.58, SD 0.13; 95% CI 0.01-0.58; P=.01). Mean cross-classification into exact agreement plus adjacent was 85% for nutrient intake and 83% for food groups. Bland-Altman plots showed moderate agreement for energy-adjusted macronutrient intakes. Conclusions: The results indicate that the web-based EatWellQ8 FFQ is reproducible for assessing nutrient and food group intake and has moderate agreement compared with a PFFQ and a 4-day WFR for measuring energy and nutrient intakes.

With widespread use of the internet, lifestyle and dietary data collection can now be facilitated using online questionnaires as opposed to paper versions. We have developed a graphical food frequency assessment app (eNutri), which is able to assess dietary intake using a validated food frequency questionnaire (FFQ) and provide personalised nutrition advice. FFQ user acceptance and evaluation have not been investigated extensively and only a few studies involving user acceptance of nutrition assessment and advice apps by older adults are published. A formative study with 20 participants (including n = 10 ≥60 years) assessed the suitability of this app for adults and investigated improvements to its usability. The outcomes of this formative study were applied to the final version of the application, which was deployed in an online study (EatWellUK) with 324 participants (including n = 53 ≥60 years) in the UK, using different devices (smartphones, tablets and laptops/desktops). Completion times were based on browser timestamps and usability was measured using the System Usability Scale (SUS), scoring between 0 and 100. Products with a SUS score higher than 70 are considered to be good. In the EatWellUK study, SUS score median (n = 322) was 77.5 (IQR 15.0). Out of the 322 SUS questionnaire completions, 321 device screen sizes were detected by the app. Grouped by device screen size, small (n = 92), medium (n = 38) and large (n = 191) screens received median SUS scores of 77.5 (IQR 15.0), 75.0 (IQR 19.4) and 77.5 (IQR 16.25), respectively. The median SUS scores from younger (n = 268) and older participants (n = 53) were the same. The FFQ contained 157 food items, and the mean completion time was 13.1 minutes (95% CI 12.6–13.7 minutes). Small, medium and large screen devices resulted in completion times of 11.7 minutes (95% CI 10.9–12.6 minutes), 14.4 minutes (95% CI 12.9–15.9 minutes) and 13.6 minutes (95% CI 12.8–14.3 minutes), respectively. The overall median SUS score of 77.5 and overall mean completion time of 13.3 minutes indicate good overall usability, and equally, comparable SUS scores and completion times across small, medium and large screen sizes indicates good usability across devices. This work is a step toward the promotion of wider uptake of online apps that can provide online dietary intake assessment at-scale, with the aim of addressing pressing epidemiological challenges.

Nutrition apps have great potential to support people to improve their diets, but few apps give automated validated personalised nutrition advice. A web app capable of delivering automated personalised food-based nutrition advice (eNutri) was developed. The aims of this study were to i) evaluate and optimise the personalised nutrition report provided by the app and ii) compare the personalised food-based advice with nutrition professionals’ standards to aid validation. A study with nutrition professionals (NP) compared the advice provided by the app against professional Registered Dietitians (RD) (n = 16) and Registered Nutritionists (RN) (n = 16) standards. Each NP received two pre-defined scenarios, comprising an individual’s characteristics and dietary intake based on an analysis of a food frequency questionnaire, along with the nutrition food-based advice that was automatically generated by the app for that individual. NPs were asked to use their professional judgment to consider the scenario, provide their three most relevant recommendations for that individual, then consider the app’s advice and rate their level of agreement via 5-star scales (with 5 as complete agreement). NPs were also asked to comment on the eNutri recommendations, scores generated and overall impression. The mean scores for the appropriateness, relevance and suitability of the eNutri diet messages were 3.5, 3.3 and 3.3 respectively.