The eButton takes frontal images at 4s intervals throughout the day. A three-dimensional manually administered wire mesh procedure has been developed to quantify portion sizes from the two-dimensional images. The present paper reports a test of the inter-rater reliability and validity of use of the wire mesh procedure.

Seventeen foods of diverse shapes and sizes served on plates, bowls and cups were selected to rigorously test the portion assessment procedure. A dietitian not involved in inter-rater reliability assessment used standard cups to independently measure the quantities of foods to generate the ‘true’ value for a total of seventy-five ‘served’ and seventy-five smaller ‘left’ images with diverse portion sizes.

The images appeared on the computer to which the digital wire meshes were applied.

Two dietitians and three engineers independently estimated portion size of the larger (‘served’) and smaller (‘left’) images for the same foods.

The engineers had higher reliability and validity than the dietitians. The dietitians had lower reliabilities and validities for the smaller more irregular images, but the engineers did not, suggesting training could overcome this limitation. The lower reliabilities and validities for foods served in bowls, compared with plates, suggest difficulties with the curved nature of the bowls.

The wire mesh procedure is an important step forward in quantifying portion size, which has been subject to substantial self-report error. Improved training procedures are needed to overcome the identified problems.

For dietary assessment, mobile devices with a camera can be used as an alternative to hand-written paper records. The Nutritional Tracking Information Smartphone (Nutris-Phone) study aimed to examine relative validity and feasibility of a photo-based dietary record in everyday life.

Parallel to the photo-based technique, a weighed record was performed. Participant satisfaction was assessed by questionnaire. A trained nutrition scientist evaluated portion sizes and nutrient content was calculated (DGExpert). Spearman correlation and Bland–Altman analyses were applied.

Healthy, non-pregnant volunteers (≥18 years) without intent to lose weight recruited at Ulm University, Germany.

Sixty-six participants (36 % males, median age 22·0 (interquartile range 20·0–25·0) years) took pictures of foods and beverages consumed with a commercially available mobile phone.

Significant correlation between the photo-based and weighed record was observed: energy (r=0·991), carbohydrate (r=0·980), fat (r=0·972), protein (r=0·988), fibre (r=0·941). Bland–Altman analyses indicated comparable means and acceptable 95 % limits of agreement (energy: −345·2 to 302·9 kJ (−82·5 to 72·4 kcal); carbohydrate: −15·2 to 13·1 g; fat: −6·4 to 6·4 g; protein: −5·9 to 5·6 g; fibre: −2·7 to 2·5 g). However, with increasing intake level, underestimation by the digital method was present (except for fat, all P<0·01). Over 80 % of participants were satisfied with the photo-based record. In nearly 90 %, technical implementation was without major problems.

Compared with a weighed record, the photo-based dietary record seems to be valid, feasible and user-friendly to estimate energy, macronutrient and fibre intakes, although a systematic bias with increasing levels of intake should be kept in mind.

To develop an artificial intelligence (AI)-based algorithm which can automatically detect food items from images acquired by an egocentric wearable camera for dietary assessment.

To study human diet and lifestyle, large sets of egocentric images were acquired using a wearable device, called eButton, from free-living individuals. Three thousand nine hundred images containing real-world activities, which formed eButton data set 1, were manually selected from thirty subjects. eButton data set 2 contained 29 515 images acquired from a research participant in a week-long unrestricted recording. They included both food- and non-food-related real-life activities, such as dining at both home and restaurants, cooking, shopping, gardening, housekeeping chores, taking classes, gym exercise, etc. All images in these data sets were classified as food/non-food images based on their tags generated by a convolutional neural network.

A cross data-set test was conducted on eButton data set 1. The overall accuracy of food detection was 91·5 and 86·4 %, respectively, when one-half of data set 1 was used for training and the other half for testing. For eButton data set 2, 74·0 % sensitivity and 87·0 % specificity were obtained if both ‘food’ and ‘drink’ were considered as food images. Alternatively, if only ‘food’ items were considered, the sensitivity and specificity reached 85·0 and 85·8 %, respectively.

The AI technology can automatically detect foods from low-quality, wearable camera-acquired real-world egocentric images with reasonable accuracy, reducing both the burden of data processing and privacy concerns.

Current approaches to food volume estimation require the person to carry a fiducial marker (e.g. a checkerboard card), to be placed next to the food before taking a picture. This procedure is inconvenient and post-processing of the food picture is time-consuming and sometimes inaccurate. These problems keep people from using the smartphone for self-administered dietary assessment. The current bioengineering study presents a novel smartphone-based imaging approach to table-side estimation of food volume which overcomes current limitations.

We present a new method for food volume estimation without a fiducial marker. Our mathematical model indicates that, using a special picture-taking strategy, the smartphone-based imaging system can be calibrated adequately if the physical length of the smartphone and the output of the motion sensor within the device are known. We also present and test a new virtual reality method for food volume estimation using the International Food Unit™ and a training process for error control.

Our pilot study, with sixty-nine participants and fifteen foods, indicates that the fiducial-marker-free approach is valid and that the training improves estimation accuracy significantly (P<0·05) for all but one food (egg, P>0·05).

Elimination of a fiducial marker and application of virtual reality, the International Food Unit™ and an automated training allowed quick food volume estimation and control of the estimation error. The estimated volume could be used to search a nutrient database and determine energy and nutrients in the diet.

The present study tested the combination of an established and a validated food-choice research method (the ‘fake food buffet’) with a new food-matching technology to automate the data collection and analysis.

The methodology combines fake-food image recognition using deep learning and food matching and standardization based on natural language processing. The former is specific because it uses a single deep learning network to perform both the segmentation and the classification at the pixel level of the image. To assess its performance, measures based on the standard pixel accuracy and Intersection over Union were applied. Food matching firstly describes each of the recognized food items in the image and then matches the food items with their compositional data, considering both their food names and their descriptors.

The final accuracy of the deep learning model trained on fake-food images acquired by 124 study participants and providing fifty-five food classes was 92·18 %, while the food matching was performed with a classification accuracy of 93 %.

The present findings are a step towards automating dietary assessment and food-choice research. The methodology outperforms other approaches in pixel accuracy, and since it is the first automatic solution for recognizing the images of fake foods, the results could be used as a baseline for possible future studies. As the approach enables a semi-automatic description of recognized food items (e.g. with respect to FoodEx2), these can be linked to any food composition database that applies the same classification and description system.

Optimal nutrition for hospital patients is crucial and routine monitoring of patients’ nutrient intake is imperative. However, personalised monitoring and customised intervention using traditional methods is challenging and labour-intensive, consequently it is often neglected in hospital settings. The present pilot study aimed to examine the reliability and validity of the Dietary Intake Monitoring System (DIMS) against the weighed food method (WFM).

The DIMS 2.0 is composed of an integrated digital camera, weighing scale, radio-frequency identification sensor and WIFI connection for real-time image and weight dietary data acquisition and analysis. The DIMS equipment was used to collect data for a paired set of meals both before and after meal consumption at lunchtime.

Odense University Hospital, Denmark.

Photos and weights of seventeen patient meals were captured.

The results showed a significant correlation between DIMS and WFM for energy (r=0·99, P<0·01) and protein intake (r=0·98, P<0·01). Intraclass correlation coefficients (ICC) revealed a high degree of agreement among the four non-trained assessors for estimates of portion size of each food item before (0·88, P<0·01) and after consumption (0·99, P<0·01). The ICC for energy and protein intake were 0·99 (P<0·01) and 0·99 (P<0·01), respectively. Bland–Altman plots revealed no systematic bias.

Considering the huge benefits associated with routine monitoring, technological advances have made it possible to develop a novel, easy-to-use DIMS that, according to the findings, is a valid alternative for use in hospital settings.

To verify the reliability of information, the sources of information used and the user opinions of the free mobile applications (apps) with nutritional information available in Brazil.

Descriptive, cross-sectional study.

We evaluated the content about nutrition of free apps available on the App Store of iPhone 5S with software iOS 8.4.1 and on the Play Store of the Android platform, version 2.3.6. For this, we compared the nutrition information provided by the app with (i) the Brazilian Food Composition Table (TACO), of 2011; (ii) food composition table: support for nutritional decision, of 2002; and (iii) the National Study of Family Expenditure: food composition tables, of 1999. The evaluation included the description and quantity of macro- and micronutrients in foods. In addition, we evaluated the trustworthiness of information about food energy values and analysed the comments and ratings made by users.

Mobile apps related to nutrition.

We assessed sixteen apps for mobile devices. Considering the foods selected (a basic Brazilian food basket for the month of August 2015), the apps presented partially adequate or inadequate information about food composition (macro- and micronutrients). The adequacy of the food energy values ranged from 0 to 57·1 %. Despite this, the apps received positive ratings by users.

The mobile apps about nutrition currently available and evaluated in the present study in Brazil are not useful for nutritional guidance because most of them are not based on reliable sources of information.

The quality of labelled food product databases underlying popular diet applications (apps) with barcode scanners was investigated.

Product identification rates for the scanned products and the availability and accuracy of nutrient values were calculated.

One hundred food products were selected from the two largest supermarket chains in the Netherlands. Using the barcode scanners of the selected apps, the products were scanned and the results recorded as food diary entries. The collected data were exported.

Seven diet apps with barcode scanner and food recording feature were selected from the Google Play and Apple app stores.

Energy values were available for 99 % of the scanned products, of which on average 79 % deviated not more than 5 % from the true value. MyFitnessPal provided values for sixteen nutrients, while Virtuagym Food and Yazio provided values for only four nutrients. MyFitnessPal also showed the largest percentage of correctly identified products (i.e. 96 %) and SparkPeople the smallest (i.e. 5 %). The accuracy of the provided nutrient values varied greatly between apps and nutrients.

While energy was the most consistently and accurately reported value, the availability and accuracy of other values varied greatly between apps. Whereas popular diet apps with barcode scanners might be valuable tools for dietary assessments on the product and energy level, they appear less suitable for assessments on the nutrient level. The presence of user-generated database entries implies that the availability of food products might vary depending on the size and diversity of an app’s user base.

To assess the nutritional status of folate and vitamin B12 with anaemia in young children.

A cross-sectional study was conducted at the primary health-care centres of four Brazilian cities. Folate and vitamin B12 were assessed by fluoroimmunoassay. Multilevel Poisson regression models were used to explore the association of folate and vitamin B12 status in relation to anaemia in young children.

Brazil.

Children (n 460) aged 11 to 15 months.

The median (interquartile range) of serum folate was 39·7 (28·8–55·3) nmol/l and only four children presented with folate deficiency (<10 nmol/l). Surprisingly, 30·9 % of children presented with serum folate concentrations above the upper limit of detectable values by the commercial kit used for analysis. The frequency of vitamin B12 deficiency (<148 pmol/l) was 15 % and it was inversely associated with the highest tertile of serum folate concentrations (P<0·001). Having high serum folate concentration (≥50·1 nmol/l) and vitamin B12≥148 pmol/l was associated with lower frequency of anaemia in these children (prevalence ratio=0·53; 95% CI 0·30, 0·92).

High frequency of elevated serum concentration of folate was found among young Brazilian children and 15 % of them had vitamin B12 deficiency. The combination of high serum folate and normal vitamin B12 status was associated with a lower frequency of anaemia in these children. Improvements in the current strategies to promote healthy food-based complementary feeding along with prevention and control of micronutrient deficiencies are recommended to improve children’s health.

To assess iodine status and its effects on maternal thyroid function throughout pregnancy.

In the present prospective cohort study, three urinary samples were requested for urinary iodine concentration (UIC) determinations in both the first and third gestational trimesters. Serum thyrotropin (TSH) and free thyroxine (FT4) were analysed in both trimesters and thyroid antibodies were assessed once.

Rio de Janeiro, Brazil.

First-trimester pregnant women (n 243), of whom 100 were re-evaluated during the third trimester.

Iodine sufficiency was found in the studied population (median UIC=216·7 µg/l). The first- and third-trimester median UIC was 221·0 and 208·0 µg/l, respectively. TSH levels (mean (sd)) were higher in the third trimester (1·08 (0·67) v. 1·67 (0·86) mIU/l; P<0·001), while FT4 levels decreased significantly (1·18 (0·16) v. 0·88 (0·12) ng/dl; P<0·001), regardless the presence of iodine deficiency (UIC<150 µg/l) or circulating thyroid antibodies. UIC correlated (β; 95% CI) independently and negatively with age (–0·43; –0·71, –0·17) and positively with multiparity (0·15; 0·02, 0·28) and BMI (0·25; 0·00, 0·50). Furthermore, median UIC per pregnant woman tended to correlate positively with TSH (0·07; –0·01, 0·14). Women with median UIC≥250 µg/l and at least one sample ≥500 µg/l throughout pregnancy had a higher risk of subclinical hypothyroidism (OR=6·6; 95% CI 1·2, 37·4).

In this cohort with adequate iodine status during pregnancy, excessive UIC was associated with an increased risk of subclinical hypothyroidism.

The aim of the present study was to compare selected obesity indicators with comprehensive health status.

The study employed a pooled cross-sectional design.

BMI, waist circumference, waist-to-height ratio (WHtR) and body fat percentage were considered as indirect obesity indicators. The Edmonton Obesity Staging System (EOSS) was used as a composite indicator to comprehensively reflect obesity-related co-morbidities. Cohen’s κ coefficient was used to evaluate inter-measurement agreement for obesity. Conformity of indirect obesity indicators to the EOSS was assessed based on percentage agreement (proportion classified as obese and severely unhealthy as a result of obesity among the total sample), sensitivity (proportion classified as obese among individuals severely unhealthy as a result of obesity) and specificity (proportion classified as non-obese among fairly healthy individuals). Logistic regression analysis was used to identify the sociodemographic factors most strongly associated with conformity.

The study included 17338 adults from the Korea National Health and Nutrition Examination survey conducted between July 2008 and May 2011.

Level of conformity to the EOSS was highest for WHtR (60·77 %) and lowest for BMI (35·96 %). WHtR and BMI had the highest sensitivity (53·7 %) and specificity (98·4 %), respectively. Predictability of conformity was lower among men for all indirect obesity indicators.

WHtR has the greatest potential to identify individuals at risk of health problems due to obesity. Individual demographic factors must be considered in selecting the most appropriate obesity measurement.

To develop a new predictive equation for fat mass percentage (%FM) based on anthropometric measurements and to assess its ability to discriminate between obese and non-obese individuals.

Cross-sectional study.

Mexican adults.

Adults (n 275; 181 women) aged 20–63 years with BMI between 17·4 and 42·4 kg/m2.

Thirty-seven per cent of our sample was obese using %FM measured by air-displacement plethysmography (BOD POD®; Life Measurement Instruments). The fat mass was computed from the difference between weight and fat-free mass (FFM). FFM was estimated using an equation obtained previously in the study from weight, height and sex of the individuals. The %FM estimated from the obtained FFM showed a sensitivity of 90·3 (95 % CI 86·8, 93·8) % and a specificity of 58·0 (95 % CI 52·1, 63·8) % in the diagnosis of obesity. Ninety-three per cent of participants with obesity and 65 % of participants without obesity were correctly classified.

The anthropometry-based equation obtained in the present study could be used as a screening tool in clinical and epidemiological studies not only to estimate the %FM, but also to discriminate the obese condition in populations with similar characteristics to the participant sample.

n-3 PUFA during pregnancy have been subject to intense research interest, but also much controversy, creating a situation of confusion among patients and health-care providers. The present study was carried out to explore knowledge, attitude and practices regarding n-3 PUFA in two independent populations of pregnant women and gynaecologists–obstetricians in Belgium and to assess the relationship between the pregnant women’s attributes and their n-3 PUFA status measured by the omega-3 index.

Cross-sectional study. Knowledge, attitude and practices were collected by self-administered questionnaires while laboratory data were available for each pregnant woman.

During the first antenatal hospital appointment (for the pregnant women) and by email (for the gynaecologists–obstetricians).

Women in early pregnancy (n 122) and gynaecologists–obstetricians (n 67).

Marked discrepancies in perception were evidenced between the pregnant women and health-care providers. While 82 % of the women attached high importance to n-3 PUFA during pregnancy, only a third of the gynaecologists–obstetricians did. About 35 % of the women declared paying particular attention to their consumption of n-3 PUFA. After adjusting for sociodemographic characteristics, these positive dietary practices were significantly associated with higher omega-3 index (P=0·04). Overall, 43·3 % of professionals did not provide any information about n-3 PUFA to their pregnant patients and 46·3 % did not take any preventive actions.

Evidence-based guidelines, refreshment training and communication tools are needed to improve awareness and clinical practices among caregivers regarding n-3 PUFA to benefit both mothers and their children.

The current mixed-methods study explored qualitative accounts of prior childhood experiences and current contextual factors around family meals across three quantitatively informed categories of family meal frequency patterns from adolescence to parenthood: (i) ‘maintainers’ of family meals across generations; (ii) ‘starters’ of family meals in the next generation; and (iii) ‘inconsistent’ family meal patterns across generations.

Quantitative survey data collected as part of the first (1998–1999) and fourth (2015–2016) waves of the longitudinal Project EAT (Eating and Activity in Adolescents and Young Adults) study and qualitative interviews conducted with a subset (n 40) of Project EAT parent participants in 2016–2017.

Surveys were completed in school (Wave 1) and online (Wave 4); qualitative interviews were completed in-person or over the telephone.

Parents of children of pre-school age (n 40) who had also completed Project EAT surveys at Wave 1 and Wave 4.

Findings revealed salient variation within each overarching theme around family meal influences (‘early childhood experiences’, ‘influence of partner’, ‘household skills’ and ‘family priorities’) across the three intergenerational family meal patterns, in which ‘maintainers’ had numerous influences that supported the practice of family meals; ‘starters’ experienced some supports and some challenges; and ‘inconsistents’ experienced many barriers to making family meals a regular practice.

Family meal interventions should address differences in cooking and planning skills, aim to reach all adults in the home, and seek to help parents who did not eat family meals as a child develop an understanding of how and why they might start this tradition with their family.

While maternal folate deficiency has been linked to poor pregnancy outcomes such as neural tube defects, anaemia and low birth weight, the relationship between folate and preterm birth (PTB) in the context of the US post-folic acid fortification era is inconclusive. We sought to explore the relationship between maternal folate status and PTB and its subtypes, i.e. spontaneous and medically indicated PTB.

Observational study.

Boston Birth Cohort, a predominantly urban, low-income, race/ethnic minority population at a high risk for PTB.

Mother–infant dyads (n 7675) enrolled in the Boston Birth Cohort. A sub-sample (n 2313) of these dyads had maternal plasma folate samples collected 24–72 h after delivery.

Unadjusted and adjusted logistic regressions revealed an inverse relationship between the frequency of multivitamin supplement intake and PTB. Compared with less frequent use, multivitamin supplement intake 3–5 times/week (adjusted OR (aOR) = 0·78; 95 % CI 0·64, 0·96) or >5 times/week (aOR = 0·77; 95 % CI 0·64, 0·93) throughout pregnancy was associated with reduced risk of PTB. Consistently, higher plasma folate levels (highest v. lowest quartile) were associated with lower risk of PTB (aOR = 0·74; 95 % CI 0·56, 0·97). The above associations were similar among spontaneous and medically indicated PTB.

If confirmed by future studies, our findings raise the possibility that optimizing maternal folate levels across pregnancy may help to reduce the risk of PTB among the most vulnerable US population in the post-folic acid fortification era.

To assess knowledge of osteoporosis and its risk factors and to explore associations between knowledge and various sociodemographic factors in Indian adults.

Cross-sectional study. The Revised Osteoporosis Knowledge Test (OKT) was used to assess knowledge of osteoporosis. Four scores (OKT-total, range 0–32; OKT-exercise, range 0–20; OKT-nutrition, range 0–26; OKT-risk factors, range 0–14) were generated by giving 1 point to every correct answer and 0 points for incorrect or ‘not known’ answers.

Tertiary-care hospital in Pune city, India.

Adults aged 40–75 years (n 477; 234 males) enrolled through voluntary routine health checks and health camps.

Mean age of the study population was 54·6 (sd 9·5) years. Half the participants were aware of osteoporosis and could correctly define it. Women showed significantly higher median OKT-total and OKT-nutrition scores than men (P<0·05). Those with higher education and higher socio-economic status had significantly higher scores in both men and women (P<0·05). All four scores were significantly higher in both men and women who could correctly define osteoporosis (P<0·05). All four scores were significantly higher in women with a family history of osteoporosis (P<0·05) but not in men (P>0·1).

Understanding about osteoporosis and its risk factors is low in the present cohort of Indian men and women. There is need to create awareness programmes aimed at both men and women especially targeting those with lower education, lower socio-economic status and no previous exposure to osteoporosis.

To evaluate the secondary impact of a multilevel, child-focused, obesity intervention on food-related behaviours (acquisition, preparation, fruit and vegetable (FV) consumption) on youths’ primary caregivers.

B’More Healthy Communities for Kids (BHCK) group-randomized controlled trial promoted access to healthy foods and food-related behaviours through wholesaler and small store strategies, peer mentor-led nutrition education aimed at youths, and social media and text messaging targeting their adult caregivers. Measures included caregivers’ (n 516) self-reported household food acquisition frequency for FV, snacks and grocery items over 30 d, and usual FV consumption in a sub-sample of 226 caregivers via the NCI FV Screener. Hierarchical models assessed average treatment effects (ATE). Treatment-on-the-treated-effect (TTE) analyses evaluated correlation between behavioural change and exposure to BHCK. Exposure scores at post-assessment were based on self-reported viewing of BHCK materials and participating in activities.

Thirty Baltimore City low-income neighbourhoods, USA.

Adult caregivers of youths aged 9–15 years.

Of caregivers, 90·89 % were female; mean age 39·31 (sd 9·31) years. Baseline mean (sd) intake (servings/d) was 1·30 (1·69) fruits and 1·35 (1·05) vegetables. In ATE, no significant intervention effect was found on caregivers’ food-related behaviours. In TTE, each point increase in BHCK exposure score (range: 0–6·9) increased caregivers’ daily fruit consumption by 0·2 servings (0·24 (se 0·11); 95 % CI 0·04, 0·47). Caregivers reporting greater social media exposure tripled their daily fruit intake (3·16 (se 0·92); 95 % CI 1·33, 4·99) and increased their frequency of unhealthy food purchasing v. baseline.

Child-focused community-based nutrition interventions may also benefit family members’ fruit intake. Child-focused interventions should involve adult caregivers and intervention effects on family members should be assessed. Future multilevel studies should consider using social media to improve reach and engage caregiver participants.

To analyse the policy process for the South African regulation setting upper limits for salt in thirteen commonly consumed food categories, to inform future policy action for prevention of non-communicable diseases.

Semi-structured interviews (n 10) were conducted with key stakeholders from government, academia, non-governmental organisations and the food industry. Interviewees were asked about the content, context, process and actors involved in developing the regulation. Data were analysed according to Walt and Gilson’s health policy analysis triangle.

South Africa.

Key actors and stakeholders in the policy process to develop the salt regulation.

The regulation was a response to research establishing the effectiveness of food supply interventions and to a shared perception that government regulation was the quickest way to address the problem of salt overconsumption. While the regulations were developed through a consultative process, food industry stakeholders perceived the consultation as inadequate. Implementation is currently underway, supported by a health promotion programme. Monitoring and enforcement were identified as the most likely challenges due to capacity constraints.

Comprehensive mandatory salt limits are an innovative approach to food reformulation. Factors that enabled regulation included robust scientific evidence, strong political and bureaucratic leadership, and the pragmatic use of existing regulatory instruments. The main challenges identified were disagreement over the appropriate nature and extent of food industry participation, and monitoring and enforcement challenges due to capacity constraints.

Unhealthy food and drink consumption is associated with a range of physical and mental health concerns. In response, public health policies have been developed targeting a reduction in obesity in particular. In the present commentary we argue that government–industry partnerships have reduced the effectiveness of resultant policies and explore why.

Perspectives of authors.

UK.

Populations in the UK; UK Government.

Industry involvement has presented three interrelated challenges for the UK Government: (i) balancing collaboration while maintaining appropriate distance from industry stakeholders; (ii) resultant production of ‘watertight’ and effective legislation or intervention; and (iii) actual or perceived limited sanctioning or bargaining power.

Industry involvement in public health policy making has led to weak action. Support with policy implementation (rather than development) and genuine ‘buy-in’ from industry could accelerate the pace of public health improvement.