Validity of the PortionSize App for Assessing Children's Dietary Intake: Results From a Pilot Study Conducted in Controlled and Free-Living Settings.

Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Pediatr Obes . Author manuscript; available in PMC: 2026 Apr 15. Published in final edited form as: Pediatr Obes. 2026 Mar;21(3):e70102. doi: 10.1111/ijpo.70102 Search in PMC Search in PubMed View in NLM Catalog Add to search Validity of the PortionSize app for assessing children’s dietary intake: Results from a pilot study conducted in controlled and free-living settings Sanjoy Saha Sanjoy Saha 1 Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, LA, USA 2 Tennessee State University, Nashville, Tennessee, USA Find articles by Sanjoy Saha 1, 2 , Hanim E Diktas Hanim E Diktas 1 Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, LA, USA Find articles by Hanim E Diktas 1 , Chloe P Lozano Chloe P Lozano 1 Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, LA, USA 3 University of Hawaii, Cancer Center, Honolulu, HI, USA Find articles by Chloe P Lozano 1, 3 , Amanda E Staiano Amanda E Staiano 1 Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, LA, USA Find articles by Amanda E Staiano 1 , Stephanie T Broyles Stephanie T Broyles 1 Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, LA, USA Find articles by Stephanie T Broyles 1 , John W Apolzan John W Apolzan 1 Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, LA, USA Find articles by John W Apolzan 1 , Corby K Martin Corby K Martin 1 Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, LA, USA Find articles by Corby K Martin 1, * Author information Copyright and License information 1 Pennington Biomedical Research Center, Louisiana State University, Baton Rouge, LA, USA 2 Tennessee State University, Nashville, Tennessee, USA 3 University of Hawaii, Cancer Center, Honolulu, HI, USA AUTHOR CONTRIBUTIONS SS, JWA, and CKM formulated the research questions, SS, CPL, AES, SB, JWA, and CKM designed this study. SS carried out the study, and SS, HED, and STB analyzed the data. SS interpreted the findings and drafted the manuscript. All authors reviewed the manuscript. * Corresponding author: Corby K. Martin, Pennington Biomedical Research Center, Louisiana State University, 6400 Perkins Road, Baton Rouge, LA- 70808, USA, [email protected] PMC Copyright notice PMCID: PMC13077788  NIHMSID: NIHMS2162283  PMID: 41807093 The publisher's version of this article is available at Pediatr Obes Abstract Background: PortionSize is a mobile app that allows users to assess dietary intake and provide feedback in real-time. Objective: To assess the validity of the PortionSize app for estimating children’s dietary intake when assessed by the children themselves (phase 1) and relative agreement of the app when assessed by their parents (phase 2). Methods: This pilot study recruited 31 child/parent dyads. Each dyad was trained on the PortionSize app to assess dietary intake. In phase 1, children (7–12 years) assessed dietary intake in a controlled setting, and their assessments were compared with weighed meals. In phase 2, parents assessed their child’s dietary intake in a free-living setting, and the criterion measure was the digital photography of foods (DPF) method. Results: Children’s energy intake assessments using PortionSize were equivalent to weighed meals (P < 0.05, effect size: 0.22), while energy intake estimates by parents using PortionSize were not equivalent to the DPF method (P > 0.05, effect size: 0.26). The mean percent error between the measurements was 9.1% in phase 1 and 16.5% in phase 2. Conclusions: The findings suggest PortionSize may be accurate in controlled settings when used by children. In a free-living setting when used by parents, results indicated PortionSize needs improvements in methods for estimating portion size and food identification. The study findings from phase 2 reflect feasibility and relative agreement rather than validation against true intake. This pilot study focuses on initial validity, relative agreement, and user satisfaction, indicating potential for future clinical and community-based pediatric dietary and weight management applications. Keywords: food intake, food groups, energy intake, portion size, MyPlate, pediatrics, food photography 1 |. INTRODUCTION Dietary intake involves more than the estimation of quantity consumed (such as grams or kilocalories); it also includes the assessment of macro- and micronutrients, dietary quality, and dietary intake patterns. 1 Accurate assessment of dietary intake plays a pivotal role in evaluating the nutritional adequacy of children’s diet to promote healthy dietary intake as well as prevention of nutrition-related chronic diseases such as childhood obesity. 2 – 5 However, accurately assessing dietary intake in children is challenging due to their limited recall abilities, reliance on proxy reporting, and variability in eating behaviors. 6 , 7 Traditional dietary intake assessment methods (such as 24-hour dietary recall, food record, and food frequency questionnaire) are widely used to capture children’s dietary intake; however, these methods are often burdensome, subject to recall bias, or lack of precision in younger population. 3 – 5 , 8 – 11 While the digital photography of foods (DPF) method reduces participant burden, it requires trained raters to analyze food images, limiting its scalability for large studies. 12 , 13 These limitations underscore the need for innovative approaches to assess dietary intake in children and obtain immediate dietary feedback. Technology-based dietary assessment methods have the potential to reach broad populations due to increased accessibility to smartphone apps. 2 , 8 , 14 – 17 The PortionSize ™ app is an innovative smartphone app designed to assess dietary intake in real-time and provide immediate feedback on dietary intake, including MyPlate food groups. 18 Specifically, the app incorporates augmented reality (AR), which superimposes digital information (e.g., 3D portion size models) onto the real-world environment, 19 , 20 and visual comparison procedures, which allow users to compare their food portions against standardized images. 19 , 21 These tools provide an interactive and intuitive method for estimating portion sizes, reducing respondent burden while improving accuracy. 18 , 22 Visual comparison procedures and template systems are used in the PortionSize app, and it has the potential to help users correctly estimate portion size of food, and thus, energy and nutrient intake. The PortionSize app estimates food provision or selection (before meal) and food waste (after meal), thus providing an assessment of net food intake. Energy and nutrient calculations are obtained via a subset (~ 1,150) of foods from the Food and Nutrient Database for Dietary Studies (FNDDS), 23 and participants use embedded food templates to estimate portion size within the app. 18 , 22 PortionSize is the first known method that provides just-in-time feedback on dietary adherence to the USDA MyPlate food group servings. 18 , 22 Initially, the PortionSize app was developed for adults to assess their dietary intake. As part of improving the validity of the PortionSize app, a pilot study was conducted among adults to assess dietary intake in both clinical (n = 15) and free-living settings (n = 14). The results of the pilot study indicate that estimate of food intake (g) using the PortionSize app was equivalent (mean percent error: 6%) to estimate from weighed meals in a clinical setting, 18 and estimate from the DPF method in a free-living setting (mean percent error: 1.8%). 21 However, energy intake (kcal) was not equivalent to the criterion measure in both phases of the pilot study (mean percent error: 12.7% in clinical setting and 15.6% in free-living). 18 , 21 Feedback was incorporated to update the PortionSize app such as modification of templates for some foods and training instructions. Despite these improvements, the app has not been validated for use with children or by parents assessing their children’s dietary intake, leaving its accuracy in pediatric populations unknown. In this study, we adapted the PortionSize app to quantify children’s dietary intake when assessed by the children themselves and their consumed dietary intake when assessed by their parents. This pilot study had two phases. During phase 1, we tested children’s ability to estimate their intake in a single meal using PortionSize in a controlled setting, and the criterion measure was pre-prepared weighed meals, which are widely known as a gold-standard in dietary assessment. 24 As this was the first pilot study to assess children’s dietary intake using the PortionSize app, we conducted phase 1 in a controlled setting to reduce external influences (such as environmental variations) and ensure a standardized assessment environment. 18 , 25 During phase 2, among the same children, we tested the relative agreement of PortionSize when their parents used the app for two consecutive days to estimate their dietary intake compared to the DPF method in a free-living setting. Parents were involved in phase 2 because they have substantial influence on children’s dietary intake. In many families, parents are the gatekeepers for their family food environment since they decide what food is purchased, prepared and served to children, and often portion sizes of provided foods, and thus serve as a viable source of reporting their children’s dietary intake in the home environment. 26 – 28 As the app had already been piloted with adults, phase 2 was conducted in free-living settings which is a real-world environment where researchers do not have control over participants’ food consumption, 25 to capture typical eating patterns, and food choices in children’s dietary intake. 29 The criterion measure DPF was selected for phase 2 since weighed measures were not feasible due to the burden on families and the disruption this would cause in natural eating environments. 30 Although Doubly Labeled Water (DLW) is a good standard for free-living, it is cost-prohibitive and provides estimated energy intake over long periods of time; i.e., 10 to 14 days. 31 , 32 Additionally, the DPF method is an established and feasible criterion method for meal-based dietary assessment in free-living settings, which has been used in several research to assess children’s dietary intake. 9 , 11 , 30 Overall, this pilot study focused on evaluating the concurrent validity (phase 1) and relative agreement (phase 2) of the PortionSize app by comparing dietary intake estimates generated by the app to those obtained through established methods. The study hypotheses were: 1) children’s estimated energy intake using the PortionSize app would be equivalent to estimates from weighed meals in phase 1, and 2) children’s estimated energy intake by their parents using the PortionSize app would be equivalent to estimates from the DPF method in a free-living setting in phase 2. Additionally, the first secondary aim was to evaluate the equivalence of estimations of food intake by weight (g), food groups (ounce equivalent or cup equivalent), macronutrients (carbohydrate, fat, and protein), selected nutrients (saturated fat, cholesterol, dietary fiber, total sugar, and added sugar), and selected micronutrients (sodium, calcium, iron, potassium, and vitamin D) between the PortionSize app and the criterion measures. The second secondary aim was to quantify children’s and parents’ user satisfaction with the PortionSize app. 2 |. METHODS 2.1 |. Ethics approval The pilot study protocol was approved by the Institutional Review Board of Pennington Biomedical Research Center (PBRC) and was registered as a clinical trial at ClinicalTrials.gov ( NCT05587816 ) prior to participant recruitment. All procedures involving human subjects were conducted according to the Declaration of Helsinki. 33 Written consent forms from parents and assent forms from children were obtained for this study. 2.2 |. Recruitment and participants During the recruitment process, advertisements for the study were published on the PBRC website and social media. In addition to advertising on the website and social media, e-flyers for this pilot study were distributed through the Louisiana State University (LSU) AgCenter Community Nutrition Programs. We initially planned to recruit 40 dyads; however, given budget constraints and the acceptable sample size for a pilot study, we actually recruited 31 dyads for this study. A recruitment team from PBRC assessed the eligibility of dyads by having a session with parents (either mother or father) over the telephone ( Figure 1 ). The inclusion criteria of this study were: 1) children aged 7–12 years old and at least one parent or caregiver (aged 18–62 years), 2) able to use iPhone (model 6s or over), and 3) would like to comply with the study procedures. The exclusion criteria included 1) any condition or circumstance that could impede study completion, and 2) child who is very choosy eater or has a very limited food choice. Prior to participants’ visit to clinic, staff from PBRC conducted a lifestyle behavior interview via phone with parents to assess participants’ willingness and ability to complete the study procedures. Screeners who were unlikely to complete the study due to travel, household commitments, or other circumstances were excluded. FIGURE 1. Open in a new tab CONSORT diagram. Potentially eligible dyads were requested for an in-person visit at PBRC. Trained staff conducted eligibility screening for parents at the PBRC clinic. Each dyad who passed the eligibility screening at PBRC was given a subject ID number for the child and a separate subject ID number for the parent. For this study, 31 dyads were recruited between August 2022 and April 2023. Since some parents had multiple eligible children, we allowed up to two children per parent to participate in order to maximize recruitment feasibility while also minimizing excessive clustering of data from a single family. All collected data were securely stored in a password protected PBRC server. The PortionSize app complies with the Health Insurance Portability and Accountability Act. 34 Both children ($50) and parents ($75) were compensated for their successful completion of the study. 2.3 |. Procedures and measurements The study included a single visit to PBRC, followed by two days in which parents recorded their children’s dietary intake at home. 2.3.1 |. Demographics and anthropometrics During the participant’s visit to PBRC, the parent completed a short questionnaire that included the child’s date of birth, sex, race, ethnicity, education grade level, and the parent’s age, sex, race, ethnicity, income, and education level. Parent self-reported and reported on behalf of their child the sex assigned at birth as male or female. Parents also self-reported and reported on behalf of their child race or ethnicity from a list that included: non-Hispanic White, non-Hispanic Black, Hispanic, Asian or Pacific Islander, Native American (including Alaskan), biracial or multiracial, prefer not to say, do not know, or other. Trained staff recorded each child’s and parent’s height and weight twice at the PBRC clinic. These measurements were used to estimate the BMI-for-age percentile for children using CDC’s child and teen BMI calculator 35 and BMI for parents. 36 Parents were allowed to support during their child’s height and weight assessment. 2.3.2 |. Training and testing of the PortionSize app After completing the survey questionnaires and measurements, enrolled children and parents were trained on the PortionSize app for about 30–45 minutes by trained staff during their clinic visit. The first 15–20 minutes of the training session was a PowerPoint presentation for children and parents. After that, each child and parent performed a separate practice session to assess dietary intake from a sample meal (consisting of plastic food models) using the PortionSize app. The training was considered complete once the children and their parents exhibited the ability to use the PortionSize app. A detailed description of similar training and testing of PortionSize is already published. 18 , 21 , 37 Parents were allowed to assist their children in being attentive during the training session and help during the height and weight assessment. When children became proficient in using the PortionSize app, they were instructed to assess a weighed meal using the app during their visit at clinic (phase 1). Following training, parents were instructed to record at least one meal and one snack of their child for two consecutive days at home (phase 2). 2.3.3 |. Dietary assessment with the PortionSize app in a controlled setting (phase 1, performed by children) After the training session, children assessed dietary intake from a lab-based weighed meal with PortionSize without any assistance. Weighed meals, which served as the criterion measure, were pre-prepared and weighed for each child with real food items in the metabolic kitchen at PBRC. Weighed meals, a gold-standard method, 24 , 25 , 38 were used to ensure accuracy and to account for food waste, which varied considerably as described below, thereby providing a rigorous test of validity. To create the meals, children’s estimated energy requirements were calculated using equations developed by the Institute of Medicine, incorporating age, sex, height, weight, and self-reported physical activity level. 39 After estimating daily energy requirement for each child, the food provision for weighed meal included 30% of energy requirement value, representing a typical lunch/dinner for a child. 40 Five standardized menus were used in this study. Each of these meals included at least three food items. Additionally, three meals contained a calorie-containing beverage, aiming to replicate a typical American child’s meal. The menus for the weighed meals were selected from a list of commonly consumed foods from a previous study. 41 , 42 Plate waste ranged from 0% to 100%, with a mean of ~5% of the served meals to align with actual plate waste of 3% from a free-living study. 41 Conducting the study in a controlled lab setting with weighed food provision ensured precision and consistency across children’s dietary assessments while allowing for varied portion sizes. This design was crucial for assessing errors in energy estimation due to portion size. Additionally, using prepared weighed meals with real food items enabled broader inclusion criteria, as participation was not limited by food preferences. 37 , 43 Children were allowed to touch the food items but were not permitted to consume or taste the weighed meal to avoid food allergy risks. The amount ‘eaten’ by children was randomized and pictures of food selection and leftovers were taken with the PortionSize app as if they were normally consumed. Weighed meal provision and plate waste were measured separately, with dietary intake calculated by the difference. In this study, a total of 31 weighed meals were prepared, comprising five different meals with varying amounts of food items (based on energy requirements). To ensure adequate representation of each menu, we employed a restricted randomization approach. One menu was randomly assigned to each of the 31 participants, with the constraint that every menu was selected at least six times. This method preserved the randomness of assignment while maintaining balanced exposure across all menus. 44 A menu plan ( supplementary Table 1 ) with food items’ names was provided with the weighed meal to support children in typing and searching food items in PortionSize. This was necessary since children did not choose, prepare, or consume the meals. Children were informed that it was acceptable to use their own iPhone or parents’ iPhone, or a phone loaned to them by PBRC to assess dietary intake. When the children finished recording their dietary information, the PortionSize app immediately and automatically calculated dietary intake and provided instant feedback with dietary details. There was no time limit for recording meals. Parents accompanied their child in the same room; however, they were not allowed to assist the child in assessing the weighed meal using the PortionSize app. 2.3.4 |. Dietary assessment with the PortionSize app in a free-living setting (phase 2, performed by parents) After the training, parents were instructed to record at least one meal and one snack of their children per day for two consecutive days in a free-living setting. They were also instructed to maintain children’s typical eating patterns. During parents’ assessment period, reminders were sent via phone call and email when parents missed a recording, promoting them to record their child’s dietary intake for two consecutive days. They were guided to select the closest food item from the PortionSize FNDDS list when their child consumed a food not listed in PortionSize. When parents had two children (aged 7–12 years) participating in study (phase 1), they first assessed the dietary intake of the older child over two days (at least one meal and one snack per day). After completing the assessment for the older child, they then assessed the dietary intake of the younger child over the following two days. To evaluate agreement between dietary intake captured via the PortionSize app with a validated image-based method in a free-living setting, we used the DPF method as the criterion measure. The PortionSize app is designed to save the meal photos captured by users during meal assessments and automatically transfer those images wirelessly to a server managed by the research staff at PBRC. To assess dietary intake via the DPF method, trained staff members (referred to as “raters”) used the photos, taken by parents with the PortionSize app, to assess children’s dietary intake. Trained raters followed previously described and validated visual comparison procedures 4 , 5 , 45 , 46 to estimate children’s dietary intake using the DPF method. In brief, as part of the DPF method, trained raters matched each food item shown in the “before” image with the appropriate food code in the FNDDS. After that, they searched and selected a food-specific image of a standard portion size followed by comparing parents’ images with the standard portion size to assess the amounts of foods in the “before” and “after” meal images. Finally, the trained raters entered the value into a computer program built at PBRC called Food Photography Application 4 , 5 , 11 software which then estimated dietary intake based on the entered amount values with utilizing the FNDDS (version 2017–2018) dataset which contains ~7,083 food and beverage items. 23 Prior to conducting the pilot study, two raters independently analyzed some food images. The DPF method demonstrates strong interrater agreement or reliability, consistently yielding high interclass correlation coefficients (ICCs) greater than 0.9 between two raters and the details published elsewhere. 21 2.3.5 |. User satisfaction survey After completing the assessment of weighed meals using the PortionSize app (phase 1), children completed a user satisfaction survey to evaluate PortionSize on several parameters, including satisfaction with the PortionSize app, satisfaction with integrated food templates and app training, and ease of using the app. The 11-item user satisfaction survey (eight quantitative and three open ended questions) was adapted from previous studies conducted in PBRC. 4 , 5 All quantitative items were rated on a scale ranging from one to six, where one indicated “extremely dissatisfied” or “not at all,” and six indicated “extremely satisfied” or “very much.” When parents completed their children’s dietary assessment in a free-living setting (phase 2), they received an automated user satisfaction survey link to complete to evaluate PortionSize on several parameters, including ease of use, stratification, and burden. Parents who estimated dietary assessment for their second child also received a second link to complete the user satisfaction the survey. This automated user satisfaction survey was similar to the one administered in children during phase 1, which was partially adapted from previous studies conducted at PBRC. 4 , 5 2.4 |. Data analysis All statistical analyses were performed using SAS software (version 9.4; SAS Institute Inc) and IBM SPSS software (version 28.0.1; IBM Corporation). The primary outcome analysis of this study was to evaluate the equivalence of estimations of energy (kcal) intake between the PortionSize app and the criterion measure (weighed meals for phase 1 and DPF for phase 2). Furthermore, the secondary outcome analysis of this study was to evaluate the equivalence of estimations of food intake (g), food groups (ounce equivalent or cup equivalent), and specific nutrients and the criterion measures. These specific nutrients and food groups were selected because they align with dietary patterns and correspond to information provided on the nutrition fact panel. 47 The definitions of key terminology those used in this study are provided in the supplementary materials ( method section). Before conducting the analyses, the Shapiro-Wilk test was performed to confirm normality and ensure that the statistical approaches met the necessary assumptions. The test of study hypotheses was performed using the Two One-side T-Test (TOST) method and the equivalence bounds were set at ±25%; standard error estimates were corrected for non-independence of participants from the same household. Although the equivalence bounds were large, they allowed for appropriate statistical power for a pilot study and were also used in other comparable studies. 18 , 21 , 48 The TOST procedure is a statistical method designed for equivalence testing. 49 In contrast to traditional hypothesis testing, which seeks to determine whether a significant difference exists, equivalence testing evaluates whether two groups or conditions are sufficiently similar within a predefined margin or boundary. This approach is especially valuable in the validation of dietary assessment tools and other contexts where demonstrating similarity rather than identifying differences is the primary objective. 49 , 50 The p-value in an equivalence test helps to determine if the observed difference falls within a predefined range of acceptable error (the equivalence bounds). Therefore, in this case, if the p-value is less than 0.05, it indicates rejection of the null hypothesis of nonequivalence, and vice versa. We also estimated standardized effect size (Cohen’s d value) to assess the magnitude of error vs. the two criterion methods (weighed meals for phase 1 and the DPF method for phase 2). Cohen’s d values falling below 0.2 indicate a small effect size, values between 0.2 and 0.5 represent a small to medium effect size, and values equal to or exceeding 0.5 indicate a large effect size. 51 Mean percent error was calculated as: ([PortionSize mean – weighed food mean/weighed food mean] *100) for phase 1, and ([PortionSize mean – mean from the DPF method/mean from the DPF method] *100) for phase 2. Bland-Altman analyses 52 were performed to examine variations in PortionSize’s error variance across different levels of dietary intake. This analysis entailed a one-sample t-test to assess the differences between estimations with the PortionSize app and from weighed meals (phase 1) or assessments from the DPF method (phase 2). The Bland-Altman analysis involved plotting the difference in estimations of meals (PortionSize and the criterion measure) on y-axis while plotting the average of both estimates (PortionSize and the criterion measure) on x-axis. To ascertain the presence of proportional bias, the Bland-Altman analysis included a linear regression of the difference between estimates (y-axis) and average of both estimates (x-axis), with the hypothesis being that the slope did not differ from zero. An independent t-test was performed to compare energy intake between parents who assessed four meals and those who assessed three or five meals. In this pilot study with a small sample size, the significance level was set at 0.05 for all analyses. User satisfaction results from both phase 1 and phase 2 were presented primarily as frequencies and percentages. The open-ended questions from the user satisfaction survey were analyzed to identify common themes. Two researchers independently reviewed the responses, identified preliminary codes and themes, and then discussed and agreed upon the selected final themes. Open-ended responses were summarized and reported using frequencies. 2.5 |. Additional analyses to preregistration The primary outcome variable, energy intake (kcal) remains the same in the manuscript as it was indicated in the preregistration. However, several post hoc or additional secondary outcome variables are reported in this manuscript, including food intake (g), macronutrients (carbohydrate, fat, and protein), nutrients (saturated fat, cholesterol, dietary fiber, total sugar, and added sugar), and micronutrients (sodium, calcium, iron, potassium, and vitamin D). The term “food intake” was reported in preregistration; however, the term was replaced with “dietary intake” across the manuscript to ensure consistency. 3 |. RESULTS 3.1 |. Phase 1 3.1.1 |. Participants’ characteristics During phase 1, of the 31 recruited children, 30 completed the study procedures. One child did not complete the survey and measurements, thus, the final sample was 30. The mean age of the children was 9.6 (SD = 1.8) years, and the mean BMI percentile was 55.9 (SD = 32.3) ( Table 1 ). Among the children, 56.7% were boys and most of them were White (70%). TABLE 1. Background characteristics of children (phase 1) and parents who assessed children’s dietary intake (phase 2). Characteristics Mean (SD [minimum-maximum] or n (%) Children (n = 30) Age (y) 9.6 (1.8 [7–12]) Sex Girl 13 (43.3) Boy 17 (56.7) Race White 21 (70) Asian 1 (3.3) Black or African American 8 (26.7) Grade First 2 (6.7) Second 6 (20) Third 6 (20) Fourth 3 (10) Fifth 3 (10) Sixth 6 (20) Seventh 4 (13.3) Weight status Underweight 1 (3.3) Normal 20 (66.7) Overweight 5 (16.7) Obesity 4 (13.3) Height (cm) 139.1 (13.8 [114.5–170]) Weight (kg) 37.1 (15.4 [21.5–90]) BMI percentile 55.9 (32.3 [1–99.9]) Children’s parents (n = 21) a Sex Female 95.2 (20) Male 4.8 (1) Education High school diploma or GED 1 (4.8) Some college 5 (23.8) Bachelor’s degree 11 (52.4) Postgraduate degree 4 (19) Employment Full-time (40 hours/week) 10 (47.6) Part-time 8 (38.1) Unemployed 2 (9.5) Others/student 1 (4.8) Income $10,000 - $19,999 per year ($833 - $1,666 per month) 1 (4.8) $20,000 - $29,999 per year ($1,667 - $2,499 per month) 3 (14.3) $40,000 - $49,999 per year ($3,334 - $4,166 per month) 1 (4.8) $60,000 - $69,999 per year ($5,000 - $5,833 per month) 1 (4.8) $70,000 and above per year ($5,834 and above per month) 12 (57.1) Prefer not to say 3 (14.3) Open in a new tab a Among the 30 parents, 9 had a second child who participated in this study (phase 1) 3.1.2 |. Dietary assessment with the PortionSize app in a controlled setting (phase 1) Table 2 shows that the mean energy intake with the PortionSize app was 717.9 (SD = 284.6) kcal while the mean energy intake from the weighed meals (criterion measure) was 789.5 (SD = 195.4) kcal. The PortionSize estimate was equivalent at 25% equivalence bounds to the criterion variable (P = 0.022). Estimation of energy intake with the PortionSize app was underestimated by 71.6 (SD = 319.8) kcal. The Bland-Altman analysis in Figure 2 (A) suggested that error from PortionSize differed over levels of energy intake (R 2 = 0.132, P = 0.049), with underestimates at the lower energy intake levels and overestimates in the mid-range of energy intake levels. The mean percent error between the measurements was 9.1% and Cohen’s d effect size for estimating energy intake was small to medium (−0.22). TABLE 2. Comparison of energy, food intake (g), and nutrient intake estimates between PortionSize and weighed meals (phase 1, n = 30). Variable PortionSize Weighed meal Difference Equiv.± 25% Effect size Mean percent error a Linear R 2 value P-value Mean SD Mean SD Mean SD P-value Cohen’s d Energy (kcal) 717.9 284.6 789.5 195.4 −71.6 319.8 0.022 −0.22 −9.1 0.132 0.049 Food intake (g) 517.2 182.2 580.1 169 −62.9 187.6 0.013 −0.33 −10.8 0.008 0.647 Fruits (cup eq.) 0.5 0.7 0.7 0.7 −0.2 0.6 0.45 −0.29 −28.6 0.005 0.709 Vegetables (cup eq.) 0.5 0.3 0.6 0.3 −0.1 0.5 0.311 −0.23 −16.7 0.027 0.389 Grains (oz. eq.) 4.8 6 3.1 1.5 1.7 5.3 0.817 0.31 54.8 0.851 <0.001 Dairy (cup eq.) 0.7 1 0.6 0.5 0.1 1 0.459 0.13 16.7 0.402 <0.001 Protein Foods (oz. eq.) 2.3 2 3 2.5 −0.8 1.8 0.506 −0.42 −26.7 0.091 0.106 Protein (g) 31.1 14.5 40.6 20.3 −9.5 19.6 0.43 −0.49 −23.4 0.122 0.058 Total fat (g) 27.7 12.6 30.1 11.2 −2.3 14.8 0.035 −0.16 −7.6 0.015 0.523 Saturated fat (g) 7.8 3.9 9.6 3.6 −1.8 5.4 0.264 −0.33 −18.8 0.007 0.669 Cholesterol (mg) 83.2 52.3 98.8 53.1 −15.7 62.4 0.219 −0.25 −15.9 0 0.927 Carbohydrate (g) 75.3 20.7 88.7 22.9 −13.4 23.2 0.026 −0.58 −15.1 0.013 0.552 Dietary fiber (g) 5.0 2.3 6.9 2.4 −1.9 2.5 0.611 −0.74 −27.5 0.006 0.695 Total sugar (g) 26.6 14.7 27.8 9.6 −1.2 11.6 0.006 −0.10 −4.3 0.235 0.007 Added sugar (g) 2 2.3 1.9 2.1 0.2 1 0.043 0.16 10.5 0.025 0.400 Sodium (mg) 1354.3 582 1739.3 631.1 −384.9 727.9 0.356 −0.53 −22.1 0.007 0.658 Calcium (mg) 271.8 169.9 287.7 135.8 −15.9 159.3 0.032 −0.10 −5.5 0.062 0.186 Iron (mg) 4.6 2.8 4.1 1.7 0.5 2.1 0.079 0.23 12.2 0.304 0.002 Potassium (mg) 836.2 415.6 951.7 351.6 −115.5 347.2 0.034 −0.33 −12.1 0.042 0.276 Vitamin D (ug) 1.9 2.3 0.4 0.9 1.5 2.8 0.996 0.52 375 0.588 <0.001 Open in a new tab Equiv.: Equivalence; Highlighted p-value indicated significant equivalence; a Mean percent error: ([PortionSize – Weighed meals]/Weighed meals) × 100 FIGURE 2. Open in a new tab Bland-Altman analysis comparing estimation of energy (kcal) and food intake (g) between PortionSize and weighed meals. LCL: lower confidence limit; UCL: upper confidence limit. (A) There was a proportional bias in estimating energy intake (R 2 = 0.132, P = 0.049), with participants underestimating lower energy intake levels and overestimating mid-range of energy intake levels. (B) The proportional bias in estimating food intake using the PortionSize app compared to weighed food remained consistent across different intake levels (R 2 = 0.008, P = 0.647). PortionSize estimated food intake (g) was equivalent to weighed meals (P = 0.013) and the mean percent error was 10.8%. Estimation of food intake (g) with the PortionSize app was underestimated by 62.9 (SD = 187.6) g and Cohen’s d effect size for food intake (g) was small to medium (−0.33). The Bland-Altman analysis in Figure 2 (B) found that PortionSize’s error did not differ over levels of food intake (R 2 = 0.008, P = 0.647). For the PortionSize app, none of the estimates of food groups (fruits, vegetables, grains, dairy, and protein foods) were equivalent to weighed meals ( Table 2 ). The mean difference between the two measurement methods was small for fruits (−0.2 cup eq.), vegetables (−0.1 cup eq.), grains (1.7 oz. eq.), dairy (0.1 cup eq.), and protein foods (−0.8 oz. eq.). Cohen’s d effect size was small (<0.2) for estimates of dairy and small to medium for estimates of vegetables (−0.23), fruits (−0.29), grains (0.31), and protein foods (−0.42). The difference in both assessment methods in estimating fruits, vegetables, and protein food intake was consistent over different levels of food intake (P > 0.05). However, the proportional bias from PortionSize differed over levels of intake of grains (R 2 = 0.851, P < 0.001) and dairy (R 2 = 0.402, P < 0.001), with underestimates at the lower intake levels and overestimates in the mid-range of intake levels. Among the macronutrients, the mean intake estimates for total fat (P = 0.035) and carbohydrate (P = 0.026) were equivalent between PortionSize and weighed meals, whereas the mean intake estimate for protein (P = 0.43) was not. Cohen’s d effect sizes for protein, total fat, and carbohydrate were −0.49, −0.16, and −0.58, respectively. Furthermore, the difference in both assessment methods in estimating protein, total fat, and carbohydrate intake was consistent over different levels of food intake (all P > 0.05). Across the nutrients assessed, the estimates of total sugar, added sugar, calcium, and potassium with the PortionSize app were equivalent (all P < 0.05) to weighed meals. Cohen’s d effect size for all the nutrient estimates was ≤0.5 except for dietary fiber (−0.74) and sodium (−0.53). 3.1.3 |. User satisfaction survey Among the 30 children, 56.7% (17/30) rated the training as “very much” (score 1 to 6 with 6 being the best score “very much”) in terms of how well it helped them prepare to use the PortionSize app ( Table 3 ). More than half of the participants selected a rating of ‘five’ or ‘six’ for the ease of using PortionSize (both before [16/30] and after photos [21/30]) to capture information about food portion sizes on their plates. Furthermore, about 70% (21/30) of participants selected a rating of ‘five’ or ‘six’ to express their satisfaction with the feedback provided by the PortionSize app on their food assessment. For the first open-ended question, “ which aspects of PortionSize did you find the easiest to use? ”, the identified themes and response frequency were: taking photos (n = 16), using food templates (n = 5), tagging food items (n = 2), food search (n = 2), adding meal details (n = 2), dietary feedback (n = 1), leftover function (n = 1), and overall easy to use (n = 1). For the second question, “ which aspects of PortionSize did you find most difficult to use? ”, the main themes and response frequency were: using food templates (n = 11), food search (n = 7), taking photos (n = 7), adding meal details (n = 2), and leftover function (n = 2). For the third question, “ please write any additional comments you have about the pros and cons of using the PortionSize app ”, the identified themes for pros were overall ease of use (n = 1) and food search (n = 1). In contrast, the key themes for cons included making the app easier to use (n = 2), taking photos (n = 2), adding meal details (n = 1), editing meal information (n = 1), quantifying portion size estimations (n = 1), being time consuming (n = 1), and using food templates (n = 1). TABLE 3. Children’s satisfaction with the PortionSize app during phase 1 (N=30). Questions Score, n (%) 1 a 2 3 4 5 6 b 1. How much did the training help prepare you for using PortionSize? 0 (0) 0 (0) 3 (10) 3 (10) 7 (23.3) 17 (56.7) 2. Was it easy to use the PortionSize Before Photo tab to capture information about the portion size of foods on your plate before a meal was consumed? 0 (0) 7 (23.3) 5 (16.7) 2 (6.7) 11 (36.7) 5 (16.7) 3. Was it easy to use the PortionSize After Photo tab to capture information about leftovers after a meal was consumed? 0 (0) 2 (6.7) 2 (6.7) 5 (16.7) 8 (26.7) 13 (43.3) 4. Was it easy to use PortionSize to record your food intake? 0 (0) 0 (0) 7 (23.3) 7 (23.3) 9 (30) 7 (23.3) 5. How satisfied were you with using the PortionSize Before Photo tab to capture information about the portion size of foods on your plate before a meal was consumed? 0 (0) 0 (0) 5 (16.7) 6 (20) 5 (16.7) 14 (46.7) 6. How satisfied were you with using the PortionSize After Photo tab to capture information about leftovers after a meal was consumed? 0 (0) 1 (3.3) 5 (16.7) 5 (16.7) 6 (20) 13 (43.3) 7. How satisfied are you with the templates that were superimposed on food items within the PortionSize app? 1 (3.3) 2 (6.7) 6 (20) 3 (10) 8 (26.7) 10 (33.3) 8. How satisfied were you with the feedback provided by PortionSize regarding your meal totals? 0 (0) 0 (0) 2 (6.7) 7 (23.3) 6 (20) 15 (50) Open in a new tab a Score of 1 represented “not at all” for questions 1–4, and “extremely dissatisfied” for questions 5–8 b Score of 6 represented “very much” for questions 1–4, and “extremely satisfied” for questions 5–8 0 (0) indicated that no participant selected this response 3.2 |. Phase 2 3.2. 1 |. Participants’ characteristics Of the 31 children’s parents, one did not complete the survey and measurements. Among the remaining 30 parents, 21 were unique, as nine had a second child who participated in phase 1 of this study. Of the 21 parents, most had bachelor’s degree (52.4%), full-time job (47.6%), and income $70,000 and above per year (57.1%) ( Table 1 ). 3.2.2 |. Dietary assessment with the PortionSize app in a free-living setting (phase 2) Only 20 parents completed the dietary assessment, meaning that data for 28 children were included in phase 2, as one parent could have provided data for more than one child. Eight parents had two children enrolled and recorded dietary intake separately for each child on different days. Of the 28 dietary assessments completed by parents for their children, 19 included four meals (one meal and one snack per day for two days), two included five meals, and seven included three meals due to app errors (no data transmission) or phone issues. As reported in Table 4 , estimated average energy intake with the PortionSize app was 871.7 (SD = 410.3) kcal, whereas the average energy intake from the DPF method was 748.1 (SD = 274.8) kcal. PortionSize estimated energy intake was not equivalent to the DPF method (P = 0.24) and the Cohen’s d effect size for energy intake was small to medium (0.26) and the mean percent error was 16.5. The Bland-Altman analysis in Figure 3 (A) indicates that error from PortionSize differed over levels of energy intake (R 2 = 0.146, P = 0.045), with underestimates at the lower energy intake levels and overestimates in the mid to high-range of energy intake levels. We performed an independent t-test and found no significant difference (P > 0.05) in assessing energy intake between parents who assessed four meals and those who assessed three or five meals. TABLE 4. Comparison of energy, food intake (g), and nutrient intake estimates between PortionSize and the DPF method (phase 2, n = 28). PortionSize DPF method Difference Equiv. ± 25% Effect size Mean percent error a Linear R 2 value P-value Variable Mean SD Mean SD Mean SD p Cohen’s d Energy (kcal) 871.7 410.3 748.1 274.8 123.6 469.4 0.24 0.26 16.5 0.146 0.045 Food intake (g) 529.4 304.2 487.7 230 41.6 260.5 0.06 0.16 8.5 0.104 0.094 Fruits (cup eq.) 0.4 0.7 0.4 0.7 0 0.7 0.25 0.03 5.3 0.002 0.838 Vegetables (cup eq.) 0.4 0.4 0.6 0.7 −0.2 0.7 0.55 −0.24 −27.5 0.278 0.004 Grains (oz. eq.) 3.2 1.9 2.5 1.1 0.7 1.6 0.59 0.44 27.8 0.352 <0.001 Dairy (cup eq.) 0.6 0.5 0.6 0.4 0.1 0.4 0.15 0.16 10.9 0.107 0.09 Protein Foods (oz. eq.) 2.8 2.1 2.3 1.5 0.5 1.9 0.46 0.29 23.6 0.109 0.086 Protein (g) 33.5 19.2 29.8 14.6 3.6 18.4 0.14 0.19 12.2 0.088 0.126 Total fat (g) 38.6 22.4 32 16.8 6.7 27.2 0.40 0.24 20.9 0.08 0.143 Saturated fat (g) 12.1 6.9 10.1 5 2 7.5 0.37 0.27 20.2 0.106 0.091 Cholesterol (mg) 109.6 83.6 112.6 104.3 −3 87.4 0.07 −0.03 −2.7 0.071 0.172 Carbohydrate (g) 110.8 52.6 87.8 35.2 23 60.5 0.54 0.38 26.2 0.146 0.045 Dietary fiber (g) 8.1 5.8 6.7 3.7 1.4 6.1 0.39 0.22 20.1 0.182 0.024 Total sugar (g) 44.2 32.9 29.6 16 14.5 29.1 0.9 0.49 49.1 0.440 <0.001 Added sugar (g) 6.3 6.2 3.5 2.5 2.7 4.9 0.97 0.55 77.1 0.647 <0.001 Sodium (mg) 1484.3 760.9 1356.8 693.1 127.4 782.4 0.08 0.16 9.4 0.011 0.603 Calcium (mg) 330.1 263.2 290.1 139.2 40 220.6 0.22 0.18 13.8 0.398 <0.001 Iron (mg) 8.6 10.4 5.1 2.1 3.6 9.2 0.90 0.39 71.2 0.904 <0.001 Potassium (mg) 929.3 536.3 994.9 600.4 −65.5 742.3 0.10 −0.08 −6.6 0.013 0.565 Vitamin D (ug) 2.2 2.6 1.6 1.1 0.6 2.2 0.69 0.28 37.7 0.597 <0.001 Open in a new tab Equiv.: Equivalence; Highlighted p-value indicated significant equivalence; a Mean percent error: ([PortionSize – mean from the DPF method]/mean from the DPF method) × 100 FIGURE 3. Open in a new tab Bland-Altman analysis comparing estimation of energy (kcal) and food intake (g) between PortionSize and the DPF method. LCL: lower confidence limit; UCL: upper confidence limit. (A) There was a proportional bias in estimating energy intake (R 2 = 0.146, P = 0.045), with participants underestimating lower energy intake levels and overestimating mid to high-range of energy intake levels. (B) There was no proportional bias in estimating food intake (R 2 = 0.104, P = 0.094) using the PortionSize app compared to the DPF method. The average food intake (g) with the PortionSize app was 529.4 (SD = 304.2) g and was 487.7 (SD = 230) g from the DPF method. The difference between both assessments was 41.6 (SD = 260.5) g and the results from the equivalence tests found that the means were not equivalent (P = 0.06) at 25% equivalence bounds. The PortionSize app’s proportional bias did not vary across different levels of estimated food intake (R 2 = 0.104, P = 0.094, Figure 3 (B) ). Cohen’s d effect size difference was small (0.16), and the mean percent error was 8.5. The estimates of food groups (fruits, vegetables, grains, dairy, and protein foods) were not equivalent to the estimates from the DPF method (P > 0.05). Cohen’s d effect size for the difference between the two measurements was small (<0.2) for fruits and dairy. Additionally, Cohen’s d was small to medium for estimates of vegetables (−0.24), grains (0.44), and protein foods (0.29). Among the food group estimates, the mean percent error was low (5.3) for the estimations of fruit intake. Furthermore, among the five food group estimates, error did not differ over levels of food intake (all P > 0.05) for fruits, dairy, and protein foods. The estimates of protein, carbohydrate, and total fat intake (g) with the PortionSize app were not equivalent to the DPF method (P > 0.05). Cohen’s d effect size for protein was low (<0.2) and for total fat and carbohydrate was low to medium (between 0.2 and 0.5). Error in estimating protein and total fat was consistent over different levels of food intake (all P > 0.05); however, participants underestimating carbohydrate intake at lower levels and overestimates in the mid-range of intake levels (R 2 =0.146, P = 0.045). Across the assessments of nutrients intake with the PortionSize app, none of them, including saturated fat, cholesterol, dietary fiber, added sugar, and vitamin D, were equivalent (all P > 0.05) to the estimates from the DPF method. Cohen’s d effect sizes were small for the estimates of cholesterol, sodium, calcium, and potassium (<0.2) and small to medium for the estimates of dietary fiber, total sugar, iron, and vitamin D (between 0.2 and 0.5). 3.2.3 |. User satisfaction survey Results from the user satisfaction survey indicated that 60.7% (17/28) of the children’s parents rated the training as “very much” (score 1 to 6 with 6 being the best score “very much’) in terms of how well it helped them prepare to use the PortionSize app ( Table 5 ). More than 53.6% (15/28) of the children’s parents rated either ‘five’ or ‘six’ score in response to how satisfied they were with using the superimposed templates within the PortionSize to assess portion size and dietary intake as well as how satisfied they were with the feedback provided by the PortionSize app regarding the total food intake. TABLE 5. Parents’s satisfaction with the PortionSize app during phase 2 (N=28). Parents’ satisfaction with the PortionSize app during phase 2 (N=28). Questions Score, n (%) 1 a 2 3 4 5 6 b 1. How much did the training help prepare you for using PortionSize? 0 (0) 0 (0) 0 (0) 4 (14.3) 7 (25) 17 (60.7) 2. Was it easy to use the PortionSize Before Photo tab to capture information about the portion size of foods on your plate before a meal was consumed? 0 (0) 0 (0) 3 (10.7) 5 (17.9) 13 (46.4) 7 (25) 3. Was it easy to use the PortionSize After Photo tab to capture information about leftovers after a meal was consumed? 0 (0) 0 (0) 3 (10.7) 2 (7.1) 10 (35.7) 13 (46.4) 4. Was it easy to use PortionSize to record your food intake? 0 (0) 1 (3.6) 7 (25) 2 (7.1) 12 (42.9) 6 (21.4) 5. How satisfied were you with using the PortionSize Before Photo tab to capture information about the portion size of foods on your plate before a meal was consumed? 0 (0) 3 (10.7) 3 (10.7) 3 (10.7) 14 (50) 5 (17.9) 6. How satisfied were you with using the PortionSize After Photo tab to capture information about leftovers after a meal was consumed? 0 (0) 2 (7.1) 4 (14.3) 2 (7.1) 12 (42.9) 8 (28.6) 7. How satisfied are you with the templates that were superimposed on food items within the PortionSize app? 0 (0) 1 (3.6) 6 (21.4) 6 (21.4) 6 (21.4) 9 (32.1) 8. How satisfied were you with the feedback provided by PortionSize regarding your meal totals? 0 (0) 2 (7.1) 2 (7.1) 4 (14.3) 11 (39.3) 9 (32.1) Open in a new tab a Score of 1 represented “not at all” for questions 1–4, and “extremely dissatisfied” for questions 5–8 b Score of 6 represented “very much” for questions 1–4, and “extremely satisfied” for questions 5–8 0 (0) indicated that no participant selected this response For the first open-ended question of the user satisfaction survey, “ which aspects of PortionSize did you find the easiest to use? ”, the main themes and response frequency were: taking photos (n = 11), using food templates (n = 7), overall easy to use (n = 7), food search (n = 6), adding meal details (n = 4), tagging food items (n = 2), dietary feedback (n = 1), leftover functions (n = 1), and app training (n = 1). The key themes and response frequency for the second question (“ which aspects of PortionSize did you find most difficult to use? ”) were: taking photos (n = 12), food search (n = 8), using food templates (n = 5), food database (n = 4), challenging with recording meal event (n = 2), adding meal details (n = 1), editing meal information (n = 1), and recording leftovers (n = 1). For the third question, “ please write any additional comments you have about the pros and cons of using the PortionSize app , the identified themes for pros were meal reminders (n = 2), dietary feedback (n = 2), easier for older children (n = 1), leftover function (n = 1), overall easy to use (n = 1), and taking photos (n = 1). The main themes for cons included food database (n = 4), being time consuming (n = 3), the app not being appropriate for younger children (n = 2), taking photos (n = 2), need for improving app training (n = 1), app bugs (n = 1), editing meal details (n = 1), using food templates (n = 1), adding meal information (n = 1), adding a barcode scanner (n = 1), and challenges with recording meal event (n = 1). 4. |. DISCUSSION This pilot study evaluated the concurrent validity and relative agreement of the PortionSize app for accurately assessing children’s dietary intake in two phases. The study results indicate that children’s assessment of energy intake using the PortionSize app was equivalent to estimates from weighed meals in a controlled setting (phase 1), whereas parents’ assessment of their children’s energy intake using the app was not equivalent to the DPF method estimate in free-living environments (phase 2). 4.1 |. Phase 1 In a controlled setting, children’s estimated energy intake with the PortionSize app was equivalent to weighed meals, which supported the study hypothesis. PortionSize’s error was inconsistent over levels of intake, with a positive slope that reflected participants underestimated energy intake for the lower energy intake levels and overestimated for the mid to high range of energy intake levels. The PortionSize app estimated food intake (g) was equivalent to the estimation from the weighed meals. Although none of the food group estimates were equivalent to the weighed meals, Cohen’s effect size was small for dairy and small to medium for vegetables, fruits, grains, and protein foods. In summary, in this pilot study, we found statistically significant equivalent between the PortionSize app and weighed back measurements for energy, food intake (g), total fat, carbohydrate, total sugar, added sugar, calcium, and potassium intake. In contrast, estimates of food group servings (fruits, vegetables, grains, dairy, and protein foods), protein, saturated fat, cholesterol, dietary fiber, sodium, iron, and vitamin D intake obtained with the PortionSize app were not equivalent to those from weighed back in phase 1. A randomized crossover research trial (n = 42) found that adult participants underestimated energy intake using the PortionSize app compared to weighed meals, with a mean percent error of 13.3% in a laboratory controlled setting. 37 Another recent validation study suggests that PortionSize estimates by adults (n = 44) was equivalent to weighed meals in a semi-controlled free-living setting. 53 Our pilot study indicates that children underestimated energy intake compared to weighed meals, with a mean percent error of 9.1%. A study conducted in Burkina Faso found that the mean percent error of energy intake estimate from 24-hour dietary recall by children aged 10–11 years was 10.5% compared to the observed weighed records. 54 Burrows and colleagues conducted a study among children aged 8–11 years in Australia and found that the mean percent error in children’s energy intake estimates using a food frequency questionnaire was 15.2% compared with weighed food records and 6.4% compared with total energy expenditure measured by DLW. 55 With the comparable mean percent error between estimates from PortionSize app and traditional dietary assessment method, these findings provide preliminary evidence that the PortionSize app can be used for children (aged 7–12 years) to track their dietary intake in controlled settings, though further testing is needed. Additional validation studies with a larger sample size of children in controlled, semi-controlled and free-living settings are warranted to improve the accuracy of their dietary assessment using the PortionSize app. Technology-based, specifically smartphone-based dietary assessment methods, may be more engaging and appealing to children compared to traditional paper-based methods. 56 , 57 Traditional dietary intake assessment methods (such as 24-hour dietary recall, food record, and food frequency questionnaire) are widely used to capture children’s dietary intake; however, memory recall and portion size error are two limitations of traditional methods for children. 3 , 9 , 16 PortionSize was designed to help overcome these limitations by providing an estimate of dietary intake in real time and immediate feedback on specific meals and cumulative daily intake. 18 , 21 , 22 , 37 , 53 Although user satisfaction survey results indicate that most children were satisfied with the dietary feedback from the PortionSize app, these findings should be interpreted with caution, as children assessed only a single meal in a controlled setting. Through the open-ended responses, some children also reported challenges when using the PortionSize app. For example, one child reported that using templates was difficult for recording liquid food items, and two children mentioned that it was challenging to use templates to estimate portion size. Ongoing improvements to the PortionSize app, guided by the study results, including responses from open-ended questions, such as incorporating barcode scanning, may enhance children’s dietary assessment, reduce burden, and support scalable and affordable health promotion efforts. 3 , 37 , 43 4.2 |. Phase 2 When parents assessed their child’s dietary intake in a free-living setting, we found no equivalence in energy and food intake (g) estimates between the PortionSize app and the DPF method and the results did not support the hypothesis. However, Cohen’s d effect size for energy (kcal) and food intake (g) was small (≤0.2). Moreover, estimates for the five food groups, protein, total fat, saturated fat, cholesterol, carbohydrate, dietary fiber, total sugar, added sugar, sodium, calcium, iron, potassium, and vitamin D intake from the PortionSize app were also not equivalent to the assessment from the DPF method. Phase 2 was designed to assess relative agreement between the PortionSize app and the DPF method. The DPF method is a validated approach, but it is not a direct measure of food intake. Therefore, findings from phase 2 should be interpreted as evidence of relative agreement under free-living conditions rather than as assessments from a gold-standard method such as DLW. Previously, in a small pilot study (n = 14), adult participants overestimated energy intake using the PortionSize app compared to the DPF method, with a mean percent error of 15.6% in a free-living setting where participants estimated dietary intake over three full days. 21 On the other hand, we found that parents overestimated their children’s energy intake using the PortionSize app compared to the DPF method, with a mean percent error of 16.5%. Along with assessing the correct portion size of food, selection of the right food from the dataset is also important to accurately assess dietary intake with the PortionSize app. 22 Participants selected food items from a subset of the FNDDS dataset through scrolling a food list in the PortionSize app. In contrast, trained staff selected from the full FNDDS dataset to assess dietary intake as part of the DPF method. Research showed that exact food match is crucial; however, partial matches (food item from the same food group but not exact match) are common and contribute to over or under estimation of dietary intake. 22 Considering the research findings, including responses from open-ended questions, the training modules will be revised to help participants accurately find and select the foods they consume. Moreover, since the PortionSize app and its built-in FNDDS dataset were initially developed for adults, it may be necessary to revise the dataset to more accurately capture children’s dietary intake. Furthermore, parents used the early version of the PortionSize app to estimate dietary intake, and it could be a reason for experiencing app errors (no data transmission) by parents. The PortionSize app includes a ‘Forgot Meal’ feature, which we did not evaluate in this study, that allows users to record meals without a photo. This feature could enable parents to track their child’s full-day intake, including school meals. About two-thirds of the children’s parents rated their satisfaction as either five or six in response to the feedback provided by PortionSize on dietary intake. It is expected that dietary feedback on energy intake and food groups may support parents to assist children with meeting their food group recommendations. This is the initial study with the PortionSize app to estimate children’s dietary intake; however, future research incorporating improved training sessions, barcode scanning for packaged food items, and AI-supported food selection and portion size is expected to enhance the accuracy of dietary intake assessment by parents using the PortionSize app. 37 , 43 , 58 4.3 |. Strengths and limitations The pilot study has several strengths. First, the lab-based meal was weighed by the metabolic kitchen at PBRC and concealed from study participants and other research staff. Second, the energy content of each meal for each child was standardized since it was calculated based on the child’s age, sex, height, weight, and physical activity level, which ensured that children had different serving sizes even from the same meal plan. Third, the weighed meal provided to children resembled their regular lunch or dinner meal to promote the real-life experience of assessing dietary intake. Fourth, dietary assessments of children were conducted in both lab and free-living settings, which helped identify potential barriers and ways to improve the performance of the PortionSize app. Despite its many strengths, the study also had limitations. First, the study sample size was small; however, this is an appropriate sample size for a pilot study. 18 , 59 Given the pilot study design and modest sample size, study findings should be interpreted as preliminary evidence and warrant further validation studies in larger, more diverse pediatric populations. Second, most participants were white and educated, thereby limiting the generalizability of the study results. Third, nine parents contributed data for a second child, which may have introduced some non-independence in the data from phase 2, potentially limiting the generalizability of the findings. However, we corrected the standard error estimates for non-independence of participants from the same household. Future validation studies with larger and more diverse samples are warranted to strengthen the generalizability of the results. Fourth, the study focused on meal-based analysis and did not include itemized food-level measurement comparisons. Moreover, parents used a limited PortionSize FNDDS food list to estimate their children’s dietary intake in phase 2, which might have created challenges when searching for specific foods. However, our previous pilot study with adults in a controlled setting showed that energy intake estimates from the limited PortionSize FNDDS dataset were equivalent to those from the full FNDDS dataset. 22 Future studies are warranted to evaluate the accuracy of itemized food estimation among parents and to assess potential errors associated with the limited food list, using gold-standard methods such as weighed meals or DLW. Fifth, the lack of brand-specific nutrient data, such as for high-fiber-containing foods, may contribute to discrepancies between actual and estimated nutrient intake. Incorporation of barcode screening procedures with a future validation study may strengthen the study results and facilitate generalization. 5. |. Conclusions The study findings indicate that children’s assessment of energy (kcal) and food intake (g) using the PortionSize app was equivalent to weighed meals in a controlled setting. Overall, PortionSize estimates included food group servings (fruits, vegetables, grains, dairy, and protein foods), protein, saturated fat, cholesterol, dietary fiber, sodium, iron, and vitamin D intake were not equivalent to those from weighed back in phase 1. In contrast, estimates of children’s energy (kcal), food intake (g), food groups, and other nutrient intake by parents using the PortionSize app were not equivalent to the DPF method; however, effect sizes were generally modest. Phase 2 findings reflect feasibility and relative agreement rather than validation against true intake. The study findings suggest key areas for modification and improvement in the PortionSize app to improve the accuracy of assessing children’s dietary intake by themselves and by their parents. With ongoing improvements and validation studies, the app has the potential to improve children’s dietary assessment and modify eating behaviors and inform intervention strategies. This study’s findings suggest key improvements to the PortionSize app, including incorporating barcode screening for packaged foods, improved training, and simplified navigation. While this pilot study focused on concurrent validity (phase 1), relative agreement (phase 2), and user satisfaction, the findings suggest the potential future for clinical and community applications such as pediatric weight management and dietary counseling. The version of the app tested in this trial is not yet appropriate for clinical decision-making. This limitation reflects the lack of equivalence to the DPF method in free-living conditions, and observed differences in usability between parents and children. These factors underscore the need for further refinement of the PortionSize app, and research is needed to evaluate external validity, long-term feasibility, and potential clinical impacts on pediatric population. Supplementary Material Supplementary material NIHMS2162283-supplement-Supplementary_material.docx (32KB, docx) ACKNOWLEDGEMENTS The authors sincerely appreciate the children and their parents for their time and effort in this study. We extend our gratitude to Randy Ullrich and H. Raymond Allen for their valuable assistance with app development and data integration and management from the app. We are also grateful to the PBRC trained raters for assessing dietary intake using the DPF method. This research was supported by the Blue Cross and Blue Shield of Louisiana Foundation, the National Institute of Diabetes and Digestive and Kidney Diseases (R01, grant DK124558, P30 DK072476), the Louisiana Clinical and Translational Science Center (U54, grant GM104940), and the USDA National Institute of Food and Agriculture which was awarded to the author, CPL (NIFA, grant 2022-09708). HED is supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) grant F32HD116537. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the United States Department of Agriculture. Footnotes Clinical trial registration: NCT05587816 ( https://clinicaltrials.gov/study/NCT05587816?term=PortionSize%20app&rank=1 ) CONFLICT OF INTEREST STATEMENT Louisiana State University and Pennington Biomedical Research Center own the intellectual property associated with the PortionSize app. Authors CKM and JWA, who are employed by these institutions, are the inventors of the technology. DATA ACCESSIBILITY STATEMENT Data will be made available upon request. REFERENCES 1. Bailey RL. Overview of dietary assessment methods for measuring intakes of foods, beverages, and dietary supplements in research studies. Curr Opin Biotechnol. 2021;70:91–6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. Das SK, Miki AJ, Blanchard CM, Sazonov E, Gilhooly CH, Dey S, et al. Perspective: Opportunities and Challenges of Technology Tools in Dietary and Activity Assessment: Bridging Stakeholder Viewpoints. Adv Nutr. 2022;13(1):1–15. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Kouvari M, Mamalaki E, Bathrellou E, Poulimeneas D, Yannakoulia M, Panagiotakos DB. The validity of technology-based dietary assessment methods in childhood and adolescence: a systematic review. Crit Rev Food Sci Nutr. 2021;61(7):1065–80. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Martin CK, Correa JB, Han H, Allen HR, Rood JC, Champagne CM, et al. Validity of the Remote Food Photography Method (RFPM) for estimating energy and nutrient intake in near real-time. Obesity (Silver Spring). 2012;20(4):891–9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Martin CK, Han H, Coulon SM, Allen HR, Champagne CM, Anton SD. A novel method to remotely measure food intake of free-living individuals in real time: the remote food photography method. Br J Nutr. 2009;101(3):446–56. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Foster E, Adamson A. Challenges involved in measuring intake in early life: focus on methods. Proc Nutr Soc. 2014;73(2):201–9. [ DOI ] [ PubMed ] [ Google Scholar ] 7. Pérez-Rodrigo C, Artiach Escauriaza B, Artiach Escauriaza J, Polanco Allúe I. Dietary assessment in children and adolescents: issues and recommendations. Nutr Hosp. 2015;31 Suppl 3:76–83. [ DOI ] [ PubMed ] [ Google Scholar ] 8. Burrows T, Goldman S, Rollo M. A systematic review of the validity of dietary assessment methods in children when compared with the method of doubly labelled water. Eur J Clin Nutr. 2020;74(5):669–81. [ DOI ] [ PubMed ] [ Google Scholar ] 9. Martin CK, Newton RL Jr., Anton SD, Allen HR, Alfonso A, Han H, et al. Measurement of children’s food intake with digital photography and the effects of second servings upon food intake. Eat Behav. 2007;8(2):148–56. [ DOI ] [ PubMed ] [ Google Scholar ] 10. Martin CK, Nicklas T, Gunturk B, Correa JB, Allen HR, Champagne C. Measuring food intake with digital photography. J Hum Nutr Diet. 2014;27 Suppl 1(0 1):72–81. [ Google Scholar ] 11. Martin CK, Thomson JL, LeBlanc MM, Stewart TM, Newton RL Jr., Han H, et al. Children in school cafeterias select foods containing more saturated fat and energy than the Institute of Medicine recommendations. J Nutr. 2010;140(9):1653–60. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Gemming L, Utter J, Ni Mhurchu C. Image-assisted dietary assessment: a systematic review of the evidence. J Acad Nutr Diet. 2015;115(1):64–77. [ DOI ] [ PubMed ] [ Google Scholar ] 13. Whitton C, Healy JD, Collins CE, Mullan B, Rollo ME, Dhaliwal SS, et al. Accuracy and Cost-effectiveness of Technology-Assisted Dietary Assessment Comparing the Automated Self-administered Dietary Assessment Tool, Intake24, and an Image-Assisted Mobile Food Record 24-Hour Recall Relative to Observed Intake: Protocol for a Randomized Crossover Feeding Study. JMIR Res Protoc. 2021;10(12):e32891. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Burrows TL, Martin RJ, Collins CE. A systematic review of the validity of dietary assessment methods in children when compared with the method of doubly labeled water. J Am Diet Assoc. 2010;110(10):1501–10. [ DOI ] [ PubMed ] [ Google Scholar ] 15. Fazzino TL, Martin CK, Forbush K. The Remote Food Photography Method and SmartIntake App for the Assessment of Alcohol Use in Young Adults: Feasibility Study and Comparison to Standard Assessment Methodology. JMIR Mhealth Uhealth. 2018;6(9):e10460. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Foster E, Bradley J. Methodological considerations and future insights for 24-hour dietary recall assessment in children. Nutr Res. 2018;51:1–11. [ DOI ] [ PubMed ] [ Google Scholar ] 17. Villinger K, Wahl DR, Boeing H, Schupp HT, Renner B. The effectiveness of app-based mobile interventions on nutrition behaviours and nutrition-related health outcomes: A systematic review and meta-analysis. Obes Rev. 2019;20(10):1465–84. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Saha S, Lozano CP, Broyles S, Martin CK, Apolzan JW. Assessing the Initial Validity of the PortionSize App to Estimate Dietary Intake Among Adults: Pilot and Feasibility App Validation Study. JMIR Form Res. 2022;6(6):e38283. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Matsumoto Y, Okada G, Ohta M. Can Augmented Reality be Used as a Portion Size Estimation Aid Tool? A Pilot Randomized Controlled Trial. J Hum Nutr Diet. 2025;38(5):e70125. [ DOI ] [ PubMed ] [ Google Scholar ] 20. Mellos I, Probst Y. Evaluating augmented reality for ‘real life’ teaching of food portion concepts. J Hum Nutr Diet. 2022;35(6):1245–54. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Diktas HE, Lozano CP, Saha S, Broyles ST, Martin CK, Apolzan JW. Evaluating the Validity of the PortionSize Smartphone Application for Estimating Dietary Intake in Free-Living Conditions: A Pilot Study. J Nutr Educ Behav. 2024;56(9):643–52. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Saha S, Lozano CP, Broyles ST, Apolzan JW, Martin CK. Selecting foods from FNDDS when calculating food intake: Does the reference matter? J Food Compost Anal. 2024;125. [ Google Scholar ] 23. U.S. Department of Agriculture (USDA). Food and nutrient database for dietary studies (2017–2018). Accessed October 20, 2025. https://www.ars.usda.gov/northeast-area/beltsville-md-bhnrc/beltsville-human-nutrition-research-center/food-surveys-research-group/docs/fndds-download-databases/ 24. Scognamiglio U, Salvia A, Paolucci S, Garbagnati F, Caltagirone C, Musicco M. Validity of a questionnaire for the semi-quantitative evaluation of dietary intake of hospitalised patients compared to weighed records. J Hum Nutr Diet. 2012;25(6):526–33. [ DOI ] [ PubMed ] [ Google Scholar ] 25. Haws KL, Liu PJ, McFerran B, Chandon P. Examining eating: Bridging the gap between “lab eating” and “free-living eating”. J Assoc Cosum Res. 2022;7(4):403–18. [ Google Scholar ] 26. Livingstone MB, Robson PJ. Measurement of dietary intake in children. Proc Nutr Soc. 2000;59(2):279–93. [ DOI ] [ PubMed ] [ Google Scholar ] 27. Wyse R, Wolfenden L, Bisquera A. Characteristics of the home food environment that mediate immediate and sustained increases in child fruit and vegetable consumption: mediation analysis from the Healthy Habits cluster randomised controlled trial. Int J Behav Nutr Phys Act. 2015;12:118. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Benton D Role of parents in the determination of the food preferences of children and the development of obesity. Int J Obes Relat Metab Disord. 2004;28(7):858–69. [ DOI ] [ PubMed ] [ Google Scholar ] 29. National Academies of Sciences E, Medicine, Health, Medicine D, Food, Nutrition B. The National Academies Collection: Reports funded by National Institutes of Health. In: Callahan EA, editor. Approaches to Assessing Intake of Food and Dietary Supplements in Pregnant Women and Children 2 to 11 Years of Age: Proceedings of a Workshop Series. 2022.Washington (DC): National Academies Press (US). [ Google Scholar ] 30. Bekelman TA, Martin CK, Johnson SL, Glueck DH, Sauder KA, Harrall KK, et al. A comparison of the remote food photography method and the automated self-administered 24-h dietary assessment tool for measuring full-day dietary intake among school-age children. Br J Nutr. 2022;127(8):1269–78. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Buchowski MS. Doubly labeled water is a validated and verified reference standard in nutrition research. J Nutr. 2014;144(5):573–4. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Pagano AM, Williams TM. Estimating the energy expenditure of free-ranging polar bears using tri-axial accelerometers: A validation with doubly labeled water. Ecol Evol. 2019;9(7):4210–9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. J Int Bioethique. 2004;15(1):124–9. [ PubMed ] [ Google Scholar ] 34. Blumberg LJ, Nichols LM. The Health Insurance Portability and Accountability Act of 1996: summary of provisions and anticipated effects. J Med Pract Manage. 1998;14(1):13–8. [ PubMed ] [ Google Scholar ] 35. Centers for Disease Control and Prevention (CDC). Child and teen BMI calculator. Accessed October 25, 2025. https://www.cdc.gov/bmi/child-teen-calculator/index.html 36. Centers for Disease Control and Prevention (CDC). Body Mass Index (BMI). Accessed November 2, 2025. https://www.cdc.gov/bmi/index.html 37. Lozano CP, Neubig KE, Saha S, Broyles ST, Apolzan JW, Martin CK. Validity of the PortionSize application compared with that of MyFitnessPal for accurately estimating intake: a randomized crossover laboratory-based evaluation. Am J Clin Nutr. 2024;120(2):419–30. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 38. Budiningsari D, Shahar S, Manaf ZA, Susetyowati S. A simple dietary assessment tool to monitor food intake of hospitalized adult patients. J Multidiscip Healthc. 2016;9:311–22. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Trumbo P, Schlicker S, Yates AA, Poos M. Dietary reference intakes for energy, carbohydrate, fiber, fat, fatty acids, cholesterol, protein and amino acids. J Am Diet Assoc. 2002;102(11):1621–30. [ DOI ] [ PubMed ] [ Google Scholar ] 40. Food and Agriculture Organization (FAO). Nutrition guidelines and standards for school meals: a report from 33 low and middle-income countries. 2019. 41. Roe BE, Apolzan JW, Qi D, Allen HR, Martin CK. Plate waste of adults in the United States measured in free-living conditions. PLoS One. 2018;13(2):e0191813. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. Louisiana fit kids. Child nutrition programs for a healthier Louisiana. Accessed January 25, 2025. https://www.louisianafitkids.com/Home/MealPattern 43. Lozano CP, Canty EN, Saha S, Broyles ST, Beyl RA, Apolzan JW, et al. Validity of an Artificial Intelligence-Based Application to Identify Foods and Estimate Energy Intake Among Adults: A Pilot Study. Curr Dev Nutr. 2023;7(11):102009. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 44. Berger VW, Bour LJ, Carter K, Chipman JJ, Everett CC, Heussen N, et al. A roadmap to using randomization in clinical trials. BMC Med Res Methodol. 2021;21(1):168. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Williamson DA, Allen HR, Martin PD, Alfonso A, Gerald B, Hunt A. Digital photography: a new method for estimating food intake in cafeteria settings. Eat Weight Disord. 2004;9(1):24–8. [ DOI ] [ PubMed ] [ Google Scholar ] 46. Williamson DA, Allen HR, Martin PD, Alfonso AJ, Gerald B, Hunt A. Comparison of digital photography to weighed and visual estimation of portion sizes. J Am Diet Assoc. 2003;103(9):1139–45. [ DOI ] [ PubMed ] [ Google Scholar ] 47. U.S. Food and Drug Administration. Changes to the nutrition facts label 2016. Accessed October 14, 2025. https://www.fda.gov/food/food-labeling-nutrition/changes-nutrition-facts-label 48. Olafsdottir AS, Hörnell A, Hedelin M, Waling M, Gunnarsdottir I, Olsson C. Development and Validation of a Photographic Method to Use for Dietary Assessment in School Settings. PLoS One. 2016;11(10):e0163970. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 49. Lakens D Equivalence Tests: A Practical Primer for t Tests, Correlations, and Meta-Analyses. Soc Psychol Personal Sci. 2017;8(4):355–62. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 50. Greene CJ, Morland LA, Durkalski VL, Frueh BC. Noninferiority and equivalence designs: issues and implications for mental health research. J Trauma Stress. 2008;21(5):433–9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 51. Cohen J A power primer. Psychol Bull. 1992;112(1):155–9. [ DOI ] [ PubMed ] [ Google Scholar ] 52. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307–10. [ PubMed ] [ Google Scholar ] 53. Diktas HE, Lozano CP, Saha S, Broyles ST, Martin CK, Apolzan JW. Validity of the PortionSize ™ and MyFitnessPal© smartphone apps for estimating dietary intake: A randomized crossover clinical trial. J Nutr. 2025. [ Google Scholar ] 54. Arsenault JE, Moursi M, Olney DK, Becquey E, Ganaba R. Validation of 24-h dietary recall for estimating nutrient intakes and adequacy in adolescents in Burkina Faso. Matern Child Nutr. 2020;16(4):e13014. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 55. Burrows TL, Truby H, Morgan PJ, Callister R, Davies PS, Collins CE. A comparison and validation of child versus parent reporting of children’s energy intake using food frequency questionnaires versus food records: who’s an accurate reporter? Clin Nutr. 2013;32(4):613–8. [ DOI ] [ PubMed ] [ Google Scholar ] 56. Eldridge AL, Piernas C, Illner AK, Gibney MJ, Gurinović MA, de Vries JHM, et al. Evaluation of New Technology-Based Tools for Dietary Intake Assessment-An ILSI Europe Dietary Intake and Exposure Task Force Evaluation. Nutrients. 2018;11(1). [ Google Scholar ] 57. Chong KM, Chia A, Shah Budin NS, Poh BK, Jamil NA, Koh D, et al. Accuracy of a Web-Based Time-Use Diary (MEDAL) in Assessing Children’s Meal Intakes With Food Photography by Parents as Reference: Instrument Validation Study. JMIR Pediatr Parent. 2024;7:e53461. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 58. Folson GK, Bannerman B, Atadze V, Ador G, Kolt B, McCloskey P, et al. Validation of Mobile Artificial Intelligence Technology-Assisted Dietary Assessment Tool Against Weighed Records and 24-Hour Recall in Adolescent Females in Ghana. J Nutr. 2023;153(8):2328–38. [ DOI ] [ PubMed ] [ Google Scholar ] 59. Hertzog MA. Considerations in determining sample size for pilot studies. Res Nurs Health. 2008;31(2):180–91. [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary material NIHMS2162283-supplement-Supplementary_material.docx (32KB, docx) Data Availability Statement Data will be made available upon request. ACTIONS View on publisher site PDF (548.5 KB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top