Changing the default order of food items in an online grocery store may nudge healthier food choices.

Restructuring food environments, such as online grocery stores, has the potential to improve consumer health by encouraging healthier food choices. The aim of this study was to investigate whether repositioning foods within an experimental online grocery store can be used to nudge healthier choices. Specifically, we investigated whether repositioning product categories displayed on the website main page, and repositioning individual products within those categories, will influence selection. Adults residing in Australia (n = 175) were randomised to either intervention (high-fibre foods on top) or comparator condition (high-fibre foods on the bottom). Participants completed a shopping task using the experimental online grocery store, with a budget of up to AU$100 to for one person's weekly groceries. The results of this study show that the total fibre content per 100 kcal per cart (p < .001) and total fibre content per cart (p = .036) was higher in the intervention compared to comparator condition. Moreover, no statistical difference between conditions was found for the total number of fibre-source foods (p = .67), the total energy per cart (p = .17), and the total grocery price per cart (p = .70) indicating no evidence of implications for affordability. Approximately half of the participants (48%) reported that they would like to have the option to sort foods based on a specific nutrient criterion when shopping online. This study specifically showed that presenting higher-fibre products and product categories higher up on the online grocery store can increase the fibre content of customers' purchases. These findings have important implications for consumers, digital platform operators, researchers in health and food domains, and for policy makers.

A Fine-Tuned Bidirectional Encoder Representations From Transformers Model for Food Named-Entity Recognition: Algorithm Development and Validation.

BACKGROUND: Recently, food science has been garnering a lot of attention. There are many open research questions on food interactions, as one of the main environmental factors, with other health-related entities such as diseases, treatments, and drugs. In the last 2 decades, a large amount of work has been done in natural language processing and machine learning to enable biomedical information extraction. However, machine learning in food science domains remains inadequately resourced, which brings to attention the problem of developing methods for food information extraction. There are only few food semantic resources and few rule-based methods for food information extraction, which often depend on some external resources. However, an annotated corpus with food entities along with their normalization was published in 2019 by using several food semantic resources. OBJECTIVE: In this study, we investigated how the recently published bidirectional encoder representations from transformers (BERT) model, which provides state-of-the-art results in information extraction, can be fine-tuned for food information extraction. METHODS: We introduce FoodNER, which is a collection of corpus-based food named-entity recognition methods. It consists of 15 different models obtained by fine-tuning 3 pretrained BERT models on 5 groups of semantic resources: food versus nonfood entity, 2 subsets of Hansard food semantic tags, FoodOn semantic tags, and Systematized Nomenclature of Medicine Clinical Terms food semantic tags. RESULTS: All BERT models provided very promising results with 93.30% to 94.31% macro F1 scores in the task of distinguishing food versus nonfood entity, which represents the new state-of-the-art technology in food information extraction. Considering the tasks where semantic tags are predicted, all BERT models obtained very promising results once again, with their macro F1 scores ranging from 73.39% to 78.96%. CONCLUSIONS: FoodNER can be used to extract and annotate food entities in 5 different tasks: food versus nonfood entities and distinguishing food entities on the level of food groups by using the closest Hansard semantic tags, the parent Hansard semantic tags, the FoodOn semantic tags, or the Systematized Nomenclature of Medicine Clinical Terms semantic tags.

Impact of COVID-19 confinement on eating behaviours across 16 European countries: The COVIDiet cross-national study.

We aimed to evaluate the changes in eating behaviours of the adult population across 16 European countries due to the COVID-19 confinement and to evaluate whether these changes were somehow related to the severity of the containment measures applied in each country. An anonymous online self-reported questionnaire on socio-demographic characteristics, validated 14-items Mediterranean diet (MedDiet) Adherence Screener (MEDAS) as a reference of a healthy diet, eating and lifestyle behaviours prior to and during the COVID-19 confinement was used to collect data. The study included an adult population residing in 16 European countries at the time of the survey. Aggregated Stringency Index (SI) score, based on data from the Oxford COVID-19 Government Response Tracker, was calculated for each country at the time the questionnaire was distributed (range: 0-100). A total of 36,185 participants completed the questionnaire (77.6% female, 75.2% with high educational level and 42.7% aged between 21 and 35 years). In comparison to pre-confinement, a significantly higher adherence to the MedDiet during the confinement was observed across all countries (overall MEDAS score prior to- and during confinement: 5.23 ± 2.06 vs. 6.15 ± 2.06; p < 0.001), with the largest increase seen in Greece and North Macedonia. The highest adherence to MedDiet during confinement was found in Spain and Portugal (7.18 ± 1.84 and 7.34 ± 1.95, respectively). Stricter contingency restrictions seemed to lead to a significantly higher increase in the adherence to the MedDiet. The findings from this cross-sectional study could be used to inform current diet-related public health guidelines to ensure optimal nutrition is followed among the population, which in turn would help to alleviate the current public health crisis.

Mixed deep learning and natural language processing method for fake-food image recognition and standardization to help automated dietary assessment.

OBJECTIVE: 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. DESIGN: 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. RESULTS: 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 %. CONCLUSIONS: 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.

Development and validation of an electronic FFQ to assess food intake in the Slovene population.

OBJECTIVE: To develop and validate an electronic FFQ that estimates the food and nutrient intakes, especially fat intake, in a healthy, adult population from the coastal (Mediterranean) part of Slovenia. DESIGN: A new electronic FFQ was developed for a survey entitled 'A multidisciplinary approach in the treatment of obesity', conducted at the University of Primorska, and validated against a 3 d weighed food record (FR). SETTING: Coastal region of Slovenia. SUBJECTS: Our study population included eighty-five healthy adults aged 25-49 years, recruited from the local coastal region of Slovenia. Intakes of food groups, macronutrients and energy, estimated by the FR and the FFQ, were compared using correlation coefficients, cross-classification and Bland-Altman plots. RESULTS: The mean value of most nutrient intakes tended to be higher in the FFQ compared with the FR, except for carbohydrate, fibre and energy. Regression analysis demonstrated an acceptable agreement between the FFQ and FR. The FFQ was moderately correlated with the FR (0·30-0·54), and most of the correlations increased after energy adjustment and after de-attenuation. Relatively high (more than 70 %) proportions of participants were correctly classified into the same or an adjacent quartile. Bland-Altman analysis confirmed an acceptable level of agreement between the two methods. CONCLUSIONS: The electronic FFQ, developed for a healthy adult Slovene population from the coastal region, was shown to be a valid tool to assess food group and nutrient intakes, especially fat intakes, and to rank individuals by their intakes within gene-nutrient studies.

Comparison of paper- and web-based dietary records: a pilot study.

BACKGROUND/AIMS: Paper-based dietary records (Paper-DR) can be replaced by web-based dietary records (Web-DR) in both epidemiological studies and clinical practice to reduce the time and logistic burden. We aimed to compare Paper-DR and Web-DR. METHODS: We compared the matching of different food items (n = 1,103) from Paper-DR and Web-DR for energy and 48 nutrients among 16 pregnant volunteers, with DR for the same individuals matched for the same 4 days. Paper-DR were coded into the web-based version (referred to as Paper-Web-DR) independently by the same research dietitian. The Wilcoxon signed-rank test comparing mean rank differences, Spearman's ρ to measure associations and Bland-Altman limits of agreement to evaluate the level of agreement between the two dietary methods across the range of parameters were used. Volunteers also completed an evaluation questionnaire regarding the user acceptability of Paper-DR and Web-DR. RESULTS: A high correlation between Paper-DR and Web-DR was noted. There were statistically insignificant differences among 45 nutrients, except for free sugars (p < 0.001), α-linolenic acid (p = 0.041), folate (p = 0.036) and pantothenic acid (p = 0.023). Volunteers found the Paper-DR equally time-consuming as the Web-DR. The majority of the volunteers (75%) preferred the Web-DR. CONCLUSIONS: Paper-DR and Web-DR were comparable across a range of nutritional parameters, with a few exceptions. The Web-DR was more convenient for the majority and has substantial logistic and cost advantages.

NutriGreen image dataset: a collection of annotated nutrition, organic, and vegan food products.

Introduction: In this research, we introduce the NutriGreen dataset, which is a collection of images representing branded food products aimed for training segmentation models for detecting various labels on food packaging. Each image in the dataset comes with three distinct labels: one indicating its nutritional quality using the Nutri-Score, another denoting whether it is vegan or vegetarian origin with the V-label, and a third displaying the EU organic certification (BIO) logo. Methods: To create the dataset, we have used semi-automatic annotation pipeline that combines domain expert annotation and automatic annotation using a deep learning model. Results: The dataset comprises a total of 10,472 images. Among these, the Nutri-Score label is distributed across five sub-labels: Nutri-Score grade A with 1,250 images, grade B with 1,107 images, grade C with 867 images, grade D with 1,001 images, and grade E with 967 images. Additionally, there are 870 images featuring the V-Label, 2,328 images showcasing the BIO label, and 3,201 images without before-mentioned labels. Furthermore, we have fine-tuned the YOLOv5 segmentation model to demonstrate the practicality of using these annotated datasets, achieving an impressive accuracy of 94.0%. Discussion: These promising results indicate that this dataset has significant potential for training innovative systems capable of detecting food labels. Moreover, it can serve as a valuable benchmark dataset for emerging computer vision systems.

Dietary Intake of Adolescents and Alignment with Recommendations for Healthy and Sustainable Diets: Results of the SI.Menu Study.

BACKGROUND: The SI.Menu study offers the latest data on the dietary intake of Slovenian adolescents aged 10 to 17. The purpose of this study is to comprehensively assess their dietary intake (energy and nutrients) and compare their food intakes with dietary recommendations for healthy and sustainable diets. METHODS: The cross-sectional epidemiological dietary study SI.Menu (March 2017-April 2018) was conducted on a representative sample of Slovenian adolescents aged 10 to 17 years (n = 468) (230 males and 238 females). Data on dietary intake were gathered through two non-consecutive 24 h recalls, in line with the European Food Safety Authority (EFSA) EU Menu methodology. The repeated 24 h Dietary Recall (HDR) and Food Propensity Questionnaire (FPQ) data were combined to determine the usual intakes of nutrients and food groups, using the Multiple Source Method (MSM) program. RESULTS: Adolescents' diets significantly deviate from dietary recommendations, lacking vegetables, milk and dairy products, nuts and seeds, legumes, and water, while containing excessive meat (especially red meat) and high-sugar foods. This results in insufficient intake of dietary fibre, and nutrients such as vitamin D, folate, and calcium. CONCLUSIONS: The dietary intake of Slovenian adolescents does not meet healthy and sustainable diet recommendations. This study provides an important insight into the dietary habits of Slovenian adolescents that could be useful for future public health strategies.

FooDis: A food-disease relation mining pipeline.

Nowadays, it is really important and crucial to follow the new biomedical knowledge that is presented in scientific literature. To this end, Information Extraction pipelines can help to automatically extract meaningful relations from textual data that further require additional checks by domain experts. In the last two decades, a lot of work has been performed for extracting relations between phenotype and health concepts, however, the relations with food entities which are one of the most important environmental concepts have never been explored. In this study, we propose FooDis, a novel Information Extraction pipeline that employs state-of-the-art approaches in Natural Language Processing to mine abstracts of biomedical scientific papers and automatically suggests potential cause or treat relations between food and disease entities in different existing semantic resources. A comparison with already known relations indicates that the relations predicted by our pipeline match for 90% of the food-disease pairs that are common in our results and the NutriChem database, and 93% of the common pairs in the DietRx platform. The comparison also shows that the FooDis pipeline can suggest relations with high precision. The FooDis pipeline can be further used to dynamically discover new relations between food and diseases that should be checked by domain experts and further used to populate some of the existing resources used by NutriChem and DietRx.

SnackTrack-An App-Based Tool to Assess the Influence of Digital and Physical Environments on Snack Choice.

As food choices are usually processed subconsciously, both situational and food environment cues influence choice. This study developed and tested a mobile app to investigate the association between physical and digital environments on snack choices. SnackTrack was designed and used to collect data on the snack choices of 188 users in real-life settings during an 8-week feasibility trial. The app asks users to take a photo of the food they are planning to consume and to provide additional information regarding the physical environment and context in which this food was eaten. The app also displayed various user interface designs (i.e., different background images) to investigate the potential effects of images on snack choice. Preliminary results suggest that the time of snack obtainment did not have a significant effect on the healthfulness of the snacks chosen. Conversely, it was found that unhealthy background images appeared to encourage healthier snack choices. In conclusion, despite consumers having the knowledge to make healthy choices, environmental cues can alter food choices. SnackTrack, a novel tool to investigate the influence of physical and digital environments on consumers' food choices, provides possibilities for exploring what encourages (un)healthy eating behaviours.

From language models to large-scale food and biomedical knowledge graphs.

Knowledge about the interactions between dietary and biomedical factors is scattered throughout uncountable research articles in an unstructured form (e.g., text, images, etc.) and requires automatic structuring so that it can be provided to medical professionals in a suitable format. Various biomedical knowledge graphs exist, however, they require further extension with relations between food and biomedical entities. In this study, we evaluate the performance of three state-of-the-art relation-mining pipelines (FooDis, FoodChem and ChemDis) which extract relations between food, chemical and disease entities from textual data. We perform two case studies, where relations were automatically extracted by the pipelines and validated by domain experts. The results show that the pipelines can extract relations with an average precision around 70%, making new discoveries available to domain experts with reduced human effort, since the domain experts should only evaluate the results, instead of finding, and reading all new scientific papers.

Water Wins, Communication Matters: School-Based Intervention to Reduce Intake of Sugar-Sweetened Beverages and Increase Intake of Water.

We compared three interventions designed for reducing the consumption of sugar-sweetened beverages (SSBs) aimed at decreasing the risk of overweight and obesity among children. We included three experimental (n = 508) and one control school (n = 164) in Slovenia (672 children; 10-16 years) to evaluate interventions that influence behaviour change via environmental (E), communication (C), or combined (i.e., double) environmental and communication approaches (EC) compared to no intervention (NOI). Data of children from the 'intervention' and 'non-intervention' schools were compared before and after the interventions. The quantity of water consumed (average, mL/day) by children increased in the C and EC schools, while it decreased in the E and NOI schools. Children in the C and EC schools consumed less beverages with sugar (SSBs + fruit juices), and sweet beverages (beverages with: sugar, low-calorie and/or noncaloric sweeteners) but consumed more juices. The awareness about the health risks of SSB consumption improved among children of the 'combined intervention' EC school and was significantly different from the awareness among children of other schools (p = 0.03). A communication intervention in the school environment has more potential to reduce the intake of SSBs than a sole environmental intervention, but optimum results can be obtained when combined with environmental changes.

Is RED-S in athletes just another face of malnutrition?

BACKGROUND & AIMS: Relative energy deficiency syndrome in sport (RED-S) can impair the function of several body systems, resulting in short and long-term threats to athletes' health and performance. Research showed that these health and performance problems are often unrecognized, and the treatment is not adequate. The retrospective study presented in this paper aims to determine the prevalence of RED-S-related symptoms in a sample of Slovenian competitive athletes from various sports. METHODS: We performed retrospective research based on a database of 150 athletes, aged from 14 to 34, who had nutritional assessments as a part of their medical examination. Data were collected, refined and statistical analysis was performed. 77 women and 73 men were included; 113 were classified as young athletes (14-21 years) and 37 as elite athletes (more than 21 years). RESULTS: The majority (87%) of the athletes demonstrated at least one health-related symptom described by the RED-S-model; only 9% female and 18% male did not have any symptoms of RED-S. The number of different body systems with the compromised function was significantly higher (p < 0.001) in female athletes (2.9 ± 0.2) in comparison to male athletes (1.6 ± 0.1). For other health-related symptoms, there are statistically significant differences between young and elite athletes (p = 0.03), between female and male athletes (p = 0.02) and between young and elite female athletes (p = 0.01). When comparing groups by the number of all RED-S related symptoms, female athletes were more affected (p = 0.02). According to the RED-S CAT tool, the majority of athletes (64%) were classified in the yellow group, 7% of athletes have severe health and performance problems and fulfil criteria for the red group, and only 29% were classified in the green group. CONCLUSIONS: A high prevalence of RED-S-related symptoms in our sample competitive athletes indicates the high prevalence of nutrition-related medical problems in young and elite athletes. Therefore, it is necessary to incorporate nutritional risk screenings as a part of regular medical examinations of athletes. In addition, appropriate treatments for competitive athletes should be readily accessible, even for young athletes. It seems that the youth athlete population is the most endangered for developing malnutrition-related health problems. At the same time, we urgently need a more specific and simple nutritional screening tool that will allow us to identify athletes at nutritional risk or athletes who have RED-S.

Dietary Intakes of Slovenian Adults and Elderly: Design and Results of the National Dietary Study SI.Menu 2017/18.

Incomparable and insufficiently detailed information on dietary intakes are common challenges associated with dietary assessment methods. Being a European Union country, Slovenia is expected to conduct national food consumption studies in line with harmonised EU Menu methodology. The present study aimed to describe the methodology and protocols in the Slovenian nationally representative dietary survey SI.Menu 2017/18, and to assess population dietary habits with respect to food consumption and energy and macronutrient intakes. While the study targeted various population groups, this report is focused on adults. A representative sample of participants was randomly selected from the Central Register of Population according to sex, age classes and place of residency, following a two-stage stratified sampling procedure. Information on food consumption was collected with two non-consecutive 24-h dietary recalls using a web-based Open Platform for Clinical Nutrition (OPEN) software. Data were complemented with a food propensity questionnaire to adjust for usual intake distribution. Altogether, 364 adults (18-64 years) and 416 elderlies (65-74 years) were included in the data analyses. Study results highlighted that observed dietary patterns notably differ from food-based dietary guidelines. Typical diets are unbalanced due to high amounts of consumed meat and meat products, foods high in sugar, fat and salt, and low intake of fruits and vegetables and milk and dairy products. Consequently, the energy proportion of carbohydrates, proteins, and to some extent, free sugars and total fats, as well as intake of dietary fibre and total water deviates from the reference values. Age and sex were significantly marked by differences in dietary intakes, with particularly unfavourable trends in adults and men. Study results call for adoption of prevention and public health intervention strategies to improve dietary patterns, taking into account population group differences. In addition, all developed protocols and tools will be useful for further data collection, supporting regular dietary monitoring systems and trend analyses.

Semantics of Dairy Fermented Foods: A Microbiologist's Perspective.

Food ontologies are acquiring a central role in human nutrition, providing a standardized terminology for a proper description of intervention and observational trials. In addition to bioactive molecules, several fermented foods, particularly dairy products, provide the host with live microorganisms, thus carrying potential "genetic/functional" nutrients. To date, a proper ontology to structure and formalize the concepts used to describe fermented foods is lacking. Here we describe a semantic representation of concepts revolving around what consuming fermented foods entails, both from a technological and health point of view, focusing actions on kefir and Parmigiano Reggiano, as representatives of fresh and ripened dairy products. We included concepts related to the connection of specific microbial taxa to the dairy fermentation process, demonstrating the potential of ontologies to formalize the various gene pathways involved in raw ingredient transformation, connect them to resulting metabolites, and finally to their consequences on the fermented product, including technological, health and sensory aspects. Our work marks an improvement in the ambition of creating a harmonized semantic model for integrating different aspects of modern nutritional science. Such a model, besides formalizing a multifaceted knowledge, will be pivotal for a rich annotation of data in public repositories, as a prerequisite to generalized meta-analysis.

CafeteriaFCD Corpus: Food Consumption Data Annotated with Regard to Different Food Semantic Resources.

Besides the numerous studies in the last decade involving food and nutrition data, this domain remains low resourced. Annotated corpuses are very useful tools for researchers and experts of the domain in question, as well as for data scientists for analysis. In this paper, we present the annotation process of food consumption data (recipes) with semantic tags from different semantic resources-Hansard taxonomy, FoodOn ontology, SNOMED CT terminology and the FoodEx2 classification system. FoodBase is an annotated corpus of food entities-recipes-which includes a curated version of 1000 instances, considered a gold standard. In this study, we use the curated version of FoodBase and two different approaches for annotating-the NCBO annotator (for the FoodOn and SNOMED CT annotations) and the semi-automatic StandFood method (for the FoodEx2 annotations). The end result is a new version of the golden standard of the FoodBase corpus, called the CafeteriaFCD (Cafeteria Food Consumption Data) corpus. This corpus contains food consumption data-recipes-annotated with semantic tags from the aforementioned four different external semantic resources. With these annotations, data interoperability is achieved between five semantic resources from different domains. This resource can be further utilized for developing and training different information extraction pipelines using state-of-the-art NLP approaches for tracing knowledge about food safety applications.

Dietary Intake and Status of Vitamin B12 in Slovenian Population.

Vitamin B12 deficiency poses a health concern, especially in vulnerable populations. Dietary vitamin B12 intake was obtained by two 24 h dietary recalls and food propensity questionnaires in a representative Slovenian cross-sectional food consumption survey, SI.Menu (n = 1248 subjects; 10-74 years). For a subgroup of 280 participants, data on serum vitamin B12 were available through the Nutrihealth study. The estimated usual population-weighted mean daily vitamin B12 intakes were 6.2 µg (adults), 5.4 µg (adolescents), and 5.0 µg (elderly). Lower intakes were observed in females. Inadequate daily vitamin B12 intake (<4 µg) was detected in 37.3% of adolescents, 31.7% of adults, and 58.3% elderlies. The significant predictors for inadequate daily vitamin B12 intake were physical activity score in all age groups, sex in adolescents and adults, financial status and smoking in elderly, and employment in adults. Meat (products), followed by milk (products), made the highest vitamin B12 contribution in all age groups. In adolescents, another important vitamin B12 contributor was cereals. The mean population-weighted serum vitamin B12 levels were 322.1 pmol/L (adults) and 287.3 pmol/L (elderly). Low serum vitamin B12 concentration (<148 nmol/L) and high serum homocysteine (>15 µmol/L) were used as criteria for vitamin B12 deficiency. The highest deficiency prevalence was found in elderlies (7.0%), particularly in males (7.9%). Factors associated with high serum homocysteine were also investigated. In conclusion, although vitamin B12 status was generally not critical, additional attention should be focused particularly to the elderly.

Dietary Iron Intake and Biomarkers of Iron Status in Slovenian Population: Results of SI.Menu/Nutrihealth Study.

Inadequate iron intake and iron deficiency are recognised as a public health problem in the population at large, and particularly in specific subpopulations. Dietary iron intake was analysed using data of the national Slovenian food consumption study, SI.Menu (n = 1248 subjects; 10−74 years), while iron status was evaluated with laboratory analyses of blood haemoglobin, serum ferritin, and iron concentration in samples, collected in the Nutrihealth study (n = 280, adults). The estimated daily usual population-weighted mean iron intakes ranged from 16.0 mg in adults and the elderly to 16.7 in adolescents, and were lower in females for all three age groups. The main dietary iron sources in all the age groups were bread and bakery products, meat (products), fruit, and vegetables. The highest prevalence of haemoglobin anaemia was observed in females aged 51−64 years (6.7%). Critically depleted iron stores (ferritin concentration < 15 µg/L) were particularly found in premenopausal females (10.1%). Factors influencing low haemoglobin, ferritin, and iron intake were also investigated. We observed significant correlations between iron status with meat and fish intake, and with iron intake from meat and fish, but not with total iron intake. We can conclude that particularly premenopausal females are the most fragile population in terms of inadequate iron intake and iron deficiency, which should be considered in future research and public health strategies.

Deep Neural Networks for Image-Based Dietary Assessment.

Due to the issues and costs associated with manual dietary assessment approaches, automated solutions are required to ease and speed up the work and increase its quality. Today, automated solutions are able to record a person's dietary intake in a much simpler way, such as by taking an image with a smartphone camera. In this article, we will focus on such image-based approaches to dietary assessment. For the food image recognition problem, deep neural networks have achieved the state of the art in recent years, and we present our work in this field. In particular, we first describe the method for food and beverage image recognition using a deep neural network architecture, called NutriNet. This method, like most research done in the early days of deep learning-based food image recognition, is limited to one output per image, and therefore unsuitable for images with multiple food or beverage items. That is why approaches that perform food image segmentation are considerably more robust, as they are able to identify any number of food or beverage items in the image. We therefore also present two methods for food image segmentation - one is based on fully convolutional networks (FCNs), and the other on deep residual networks (ResNet).

Branded Foods Databases as a Tool to Support Nutrition Research and Monitoring of the Food Supply: Insights From the Slovenian Composition and Labeling Information System.

Branded foods databases are becoming very valuable not only in nutrition research but also for clinical practice, policymakers, businesses, and general population. In contrast to generic foods, branded foods are marked by rapid changes in the food supply because of reformulations, the introduction of new foods, and the removal of existing ones from the market. Also, different branded foods are available in different countries. This not only complicates the compilation of branded foods datasets but also causes such datasets to become out of date quickly. In this review, we present different approaches to the compilation of branded foods datasets, describe the history and progress of building and updating such datasets in Slovenia, and present data to support nutrition research and monitoring of the food supply. Manufacturers are key sources of information for the compilation of branded foods databases, most commonly through food labels. In Slovenia, the branded food dataset is compiled using standard food monitoring studies conducted at all major retailers. Cross-sectional studies are conducted every few years, in which the food labels of all available branded foods are photographed. Studies are conducted using the Composition and Labeling Information System (CLAS) infrastructure, composed of a smartphone application for data collection and online data extraction and management tool. We reviewed various uses of branded foods datasets. Datasets can be used to assess the nutritional composition of food in the food supply (i.e., salt, sugar content), the use of specific ingredients, for example, food additives, for nutrient profiling, and assessment of marketing techniques on food labels. Such datasets are also valuable for other studies, for example, assessing nutrient intakes in dietary surveys. Additional approaches are also being tested to keep datasets updated between food monitoring studies. A promising approach is the exploitation of crowdsourcing through the mobile application VesKajJes, which was launched in Slovenia to support consumers in making healthier dietary choices.