FoodByte: a personalized food scoring and additive risk assessment algorithm for smarter eating

Beyond Nutrient Profiling: Strengths, Limitations, and Emerging Challenges for Nutri-Score and Related Front-of-Pack Labeling Systems

Front-of-pack nutrition labeling systems have become important tools for improving consumer understanding of nutritional information and supporting healthier food choices. Among these systems, Nutri-Score has emerged as one of the most widely implemented nutrient profiling models in Europe. Based on the Food Standards Agency Nutrient Profiling System, Nutri-Score summarizes selected nutritional characteristics of foods through a simplified five-level color-coded scale. A growing body of evidence indicates that Nutri-Score improves consumers’ ability to compare products, supports healthier purchasing decisions, and may encourage product reformulation. Epidemiological studies have reported associations between dietary patterns characterized by less favorable nutrient profile scores and increased risks of chronic diseases and mortality. Despite these strengths, important scientific and conceptual questions remain. This review examines the principles of nutrient profiling and assesses how effectively food quality can be represented by a limited set of nutritional criteria. Particular attention is given to nutrient reductionism, the use of the standardized 100 g reference, food matrix effects, the relationship between nutrient profiling and food processing, the influence of hedonic drivers and health claims on consumer behavior, and the growing importance of environmental contaminants as dimensions of food quality not captured by current algorithms. Future food evaluation systems may benefit from integrating complementary information related to food structure, processing characteristics, dietary context, and emerging insights from systems nutrition.

Assessing Hospital Patient Nutrient Intake with an AI-Powered Food Recognition System – A Feasibility Study of the FlavoriaFlex solution

Adequate dietary intake is essential for positive clinical outcomes of hospitalized patients, yet monitoring food intake is labor-intensive and often subjective. AI-based food recognition could automate monitoring and assessment, but evidence in real-world hospital settings is limited. This study evaluated an AI-powered food recognition system, FlavoriaFlex, to assess its detection performance, deployment feasibility, and acceptability among dietitians. Previously validated in restaurant (F1 0.75, weight MAE 23.6 g, energy MAE 235 kcal), the system was deployed in a hospital ward for six days. A total of 133 meals were recorded; 102 had paired leftover images (235 total images). Manual annotation of 483 food segments provided ground truth for evaluating food recognition and menu mapping. Semi-structured interviews with dietitians assessed usability, perceived benefits, and clinical value. FlavoriaFlex enabled automatic estimation of item- and meal-level consumption, including weights and energy- and macronutrient contents. Overall food recognition accuracy was 94% (F1 0.76), remaining high for served meals (96.5%, F1 0.85) and robust for visually complex leftovers (89.5%, F1 0.71). Unknown/non-food segments were minimal (2.4% of leftovers; 0.27% of weight). A web dashboard delivered real-time visualizations, including energy and nutrient intake. Dietitians reported reduced cognitive burden, more objective assessment, and improved observability into patient dietary intake, while emphasizing the need for further validation and integration for clinical use. These findings demonstrate that FlavoriaFlex could be integrated into hospital workflows to provide accurate, clinically meaningful intake estimates, with AI-assisted food recognition offering an efficient, reliable approach to improving nutritional monitoring at scale.

A Web-Based Personalized Diet Recommendation System Using Decision Tree for Food Suitability Classification

Obesity and unhealthy eating patterns have become significant health concerns due to poor dietary habits and a lack of personalized nutritional guidance. Existing food recommendation systems often provide general recommendations without considering individual calorie and nutritional requirements. Therefore, this study aims to develop a web-based diet food recommendation system that integrates nutritional requirement calculations and Decision Tree-based food suitability classification. The system utilizes user information, including age, gender, weight, height, physical activity level, and diet goals, to calculate nutritional requirements through Body Mass Index (BMI), Basal Metabolic Rate (BMR) using the Mifflin-St Jeor method, and Total Daily Energy Expenditure (TDEE). A food dataset containing Indonesian foods and beverages was preprocessed and labeled using a rule-based approach based on macronutrient similarity scores. The Decision Tree algorithm was implemented to classify foods into suitable and unsuitable categories according to users’ nutritional requirements. Suitable foods were subsequently processed through a scoring mechanism and meal construction procedure to generate personalized meal plans. Experimental results showed that the Decision Tree model achieved an accuracy of 92.50%, precision of 78.26%, recall of 94.74%, and F1-score of 85.71%. System testing demonstrated that the developed features functioned properly and generated structured diet recommendations automatically. In conclusion, the proposed system can assist users in selecting foods according to their nutritional requirements and support healthier dietary planning.

The potential of Nutri-Score to differentiate foods in regulating food marketing to children

To evaluate the potential of Nutri-Score to identify foods unsuitable for marketing to children by assessing its agreement with two established nutrient profiling models: WHO-EURO 2023 and the Norwegian Regulation on Marketing (NORMA). Further, we explored the impact of additional criteria to improve agreement. Addition, we evaluated the agreement between a modified Nutri-Score, the «NewTools-score», and WHO-EURO 2023 and NORMA. Products from the Norwegian food composition table (N = 1,944) were used to explore scenarios for Nutri-Score (and NewTools-score) against WHO-EURO and NORMA. The scenarios were: 1) All products scoring C-E not permitted for marketing, 2) Same as scenario 1, but with additional criteria for trans fatty acids, non-sugar sweeteners (NSS), and added sugar for beverages. For NORMA, a third scenario was tested as an extension of scenario 2, additionally banning selected food categories from marketing. Agreement was assessed through cross classification. Under scenario 1, 53% of products were classified as Nutri-Score C-E, and 47% and 23% not permitted for marketing under WHO-EURO and NORMA, respectively. Overall agreement between Nutri-Score C-E and WHO-EURO was 84%, and 66% with NORMA. However, Nutri-Score allowed some products within unhealthy food categories, e.g., ice creams with NSS. Applying additional criteria in scenarios 2 and 3 had important effects on banning such products. Results for the NewTools-score were almost identical. The Nutri-Score and the NewTools-score may be used for identifying a high proportion of products unsuitable for marketing to children. However, some additional criteria, including a complete ban from marketing in selected food categories, seems warranted to better protect children from exposure to marketing of unhealthy products.

Discordant Associations of Food- and Nutrient-Based Dietary Inflammatory Scores With Cognitive Performance in Older Adults: A Cross-Sectional Study

Ultra-Processed Foods and Continuous Glucose Monitoring Metrics in Adults With Type 2 Diabetes

Aim: We examined associations between ultra-processed foods (UPFs) consumption and continuous glucose monitoring (CGM)-derived metrics using individual- and food-level measures of UPFs in adults with T2D. Methods: Adults with T2D (n=190) completed two 24-hour dietary recalls and wore blinded CGM devices for 14 days. Foods were classified by the NOVA system. Individual-level UPF intake was defined as the proportion of total energy or grams derived from UPFs (energy-weighted). Food-level intake was assessed as Mean UPF Energy (%) and Mean UPF Gram (%) across all reported food items (item-weighted). CGM outcomes included time in range (TIR), time above range (TAR), glucose management indicator (GMI), and coefficient of variation. Multivariable logistic regression models were adjusted for demographic, lifestyle, and clinical factors. Results: After adjustment, higher Mean UPF Energy (%) was associated with lower odds of achieving TIR >70% (OR=0.59, 95% CI: 0.38–0.92) and higher odds of TAR ≥ 25% (OR=1.62, 95% CI: 1.06–2.47) and GMI ≥7% (OR=1.56, 95% CI: 1.03–2.35). Individual-level UPF intake was not significantly associated with CGM-derived outcomes. Conclusions: Higher UPF energy content at the food level was associated with poorer CGM-derived glycemic outcomes. Food-level UPF measures may more closely reflect CGM-derived glycemic metrics than individual-level measures.