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.
Background: Mobile health (mHealth) applications are increasingly used for dietary assessment and self-monitoring, yet their validity and comprehensiveness remain uncertain. Aim: This study aimed to evaluate the quality, features, and nutritional accuracy of popular nutrition-related mobile applications available in Türkiye by comparing their outputs with standardized 3-day food records (3D-FRs). Methods: A total of 807 apps were identified in the App Store in Türkiye using the search terms 'diet' and 'food'. Seventeen applications meeting the inclusion criteria were systematically evaluated. App quality and functionality were evaluated using the Mobile Application Rating Scale (MARS). Features were categorized into dietary, tracking, insights, technical, educational, social, and artificial intelligence (AI) domains. Energy and macronutrient estimates from the apps were compared with 3D-FRs to determine validity. Summary: MyNetDiary, Fitatu, and HealthifyMe provided the broadest range of functionalities, while only five apps incorporated AI-based features. The mean MARS score for the 17 apps was 3.61/5, with the highest average in the 'functionality' domain (4.03) and the lowest in 'engagement' (3.27). Compared with 3D-FRs, 10 of the 13 apps overestimated energy intake (+327 kcal), 8 overestimated carbohydrate (+33 g), 12 overestimated protein (+23 g), and 12 overestimated fat (+19 g). The greatest discrepancies were observed in Cronometer, Diyet 7/24, Fitbit, Lifesum, and HealthifyMe, whereas MyNetDiary and FatSecret yielded estimates closest to reference records. Despite strong performance in usability and tracking features, most apps showed systematic overestimation with notable variability (≈+327 kcal; +19-33 g among overestimating apps), although findings are limited by the use of standardized menus.
H. Bayram, Arda Ozturkcan· Nutrition and Health· 0 citations
Monitoring dietary intake is a cornerstone of chronic disease management and healthy lifestyle promotion, yet it remains a persistently manual and error-prone process. This paper presents a real-time food recognition and calorie estimation system that leverages the YOLOv8 nano object-detection architecture coupled with a curated nutritional knowledge base. The proposed system is capable of identifying multiple food items in a single image or live webcam stream and immediately associating each detection with an estimated caloric value. We fine-tuned a pre-trained YOLOv8n model on a custom food dataset of 48 classes drawn from both Western and Indian cuisines and integrated the resulting detector with a Gradio-based web interface for accessible deployment. Experimental results demonstrate a mean average precision (mAP@0.5) of 91.7% on the held-out test set, an inference speed of 87 frames per second on an NVIDIA T4 GPU, and a mean absolute error (MAE) of 16.8 kcal per item across the test categories. The system outperforms prior approaches based on Faster R-CNN and SSD MobileNet on both speed and accuracy metrics. An open-source implementation is made available alongside a demonstration video, positioning this work as a practical foundation for mobile diet-tracking applications and clinical nutrition monitoring tools.
Chandramouli C G, Sangeetha Raj S· International Research Journ...· 0 citations
Objective: To systematically evaluate overall dietary quality and nutrient adequacy in post-bariatric surgery patients using the Healthy Eating Index-2015 (HEI-2015) and comparing daily intakes with Dietary Reference Values (DRVs).Material and Methods: This cross-sectional study evaluated 50 individuals who had undergone bariatric surgery, comprising 8 males and 42 females. The study included the measurement of current anthropometric data and the assessment of daily nutritional intake, determined through 3-day, 24-hour dietary recalls processed using BeBiS software. To assess dietary quality, the HEI-2015 scores were calculated, and the percentage of each nutrient intake meeting the DRVs was determined.Results: Despite significant anthropometric improvements, including a decrease in mean BMI from 41.8 ± 6.1 to 27.7 ± 4.5 kg/m², the mean HEI-2015 score was 55.6 ± 10.0, and 100% of the study population was classified as having a “poor” or “needs improvement” dietary status. Energy and protein intakes reached 28.7% and 43.4% of the recommended daily intake (DRV), respectively. The intake of critical micronutrients was extremely low particularly regarding the percentages of DRVs met for iron (32%), folate (30%), and calcium (23.6%). Macronutrient distribution revealed a high-fat preference (51.6% energy) and severe carbohydrate restriction (38.6 ± 18.3 g/day), inconsistent with recommended thresholds.Conclusion: These findings suggest that post-bariatric anthropometric improvement doesn’t consistently correspond with adequate dietary quality within this limited cohort. The integration of dietary quality monitoring with routine anthropometric assessment could enhance the care of post-bariatric patients.
BACKGROUND
Personalized meal planning by registered dietitian nutritionists (RDNs) is time-intensive. Large language models (LLMs) may automate drafting meal plans, but their nutritional accuracy in clinical practice is uncertain.
METHODS
In this proof-of-concept study, five outpatient RDNs and four LLMs (Gemini, CoPilot, ChatGPT 4.0, and customized ChatGPT 4.0) each generated 3-day meal plans for five validated clinical scenarios. Effectiveness was defined as accuracy in meeting pre-specified energy, protein, carbohydrate, fat, and sodium targets. Time to create plans and RDN comfort (self-rated confidence in nutritional accuracy and clinical appropriateness on 1-5 Likert scale) were recorded. Three independent RDNs, blinded to source, analyzed nutrient content using Nutritionist Pro. Group differences were assessed with t-test and ANOVA.
RESULTS
All LLMs and RDNs produced feasible meal plans. LLMs generated meal plans in under 1 min, whereas RDNs required a mean of 44 min per scenario. RDNs reported comfort levels ranging from 3.8 to 4.8. Across most scenarios, LLM plans delivered a smaller proportion of requested energy than RDN plans, which more consistently approached energy targets. Both groups performed similarly for the Mediterranean diet scenario. Overall, protein accuracy did not differ. However, in chronic kidney disease, LLMs undershot the guideline-based protein target, while RDNs tended to modestly exceed it. Accuracy for low-carbohydrate, fat, and sodium diets was comparable.
CONCLUSION
LLMs can rapidly generate clinically plausible meal plans but are less reliable than RDNs in achieving prescribed energy and selected macronutrient goals. Prompt precision is essential for nutrient-specific targets. A hybrid model in which RDNs refine LLM-generated drafts may leverage efficiency without sacrificing clinical accuracy.
M. Mundi, Osman Mohamed Elfadil, Danielle P. Johnson et al.· Nutrition in clinical practi...· 0 citations
Current front-of-package food-scoring systems, such as Nutri-Score, summarize nutritional quality using population-level criteria but do not account for individual health conditions, food additives, or allergen sensitivities. We developed FoodByte, a rule-based personalized food-scoring framework integrating nutrient composition, literature-informed additive-risk heuristics, allergen detection, and disease-specific scoring rules to generate individualized food ratings. FoodByte was evaluated using 50 commercially available packaged foods across five representative health profiles (250 product-profile evaluations) and compared with Nutri-Score. Differences between scoring systems were analyzed using descriptive statistics and a cross-classified linear mixed-effects model with random intercepts for food product and health profile. Sensitivity analyses and model diagnostics assessed score divergence and algorithmic robustness. FoodByte produced systematically lower scores than Nutri-Score for many processed foods, with an estimated mean difference of −10.49 points (95% CI: −14.50 to −6.48, p = 2.88 × 10⁻⁷). Disease-specific profiles showed larger score divergence than the control profile, with mean percentage differences ranging from 24.4% to 41.6%, compared with 11.6% for controls. Variance-component analysis indicated that intrinsic nutritional characteristics established the primary baseline for food evaluation, while disease-specific personalization selectively modified recommendations for foods containing condition-relevant nutrients or additives and preserved consistent recommendations for foods broadly suitable across health profiles. Sensitivity analyses demonstrated high internal stability across algorithmic perturbations, with correlations generally exceeding 0.98. These findings demonstrate that additive-aware, disease-specific scoring can systematically modify a population-level nutrient-profiling framework while remaining internally robust. Rather than uniformly re-ranking foods, FoodByte selectively personalizes recommendations where disease-specific nutritional considerations are most relevant. Because the products were purposively selected and no clinical outcomes were evaluated, the findings should be interpreted as evidence of algorithmic divergence and internal robustness rather than clinical superiority. Future studies should validate personalized food-scoring systems against clinical outcomes, expert assessments, and consumer decision-making.
Harshitha Poludas· Journal of high school scien...· 0 citations
BACKGROUND
Ultra-processed foods (UPFs) are a major component of modern diets and are associated with many adverse health outcomes. Accurate assessment of UPF consumption is essential for nutritional research.
OBJECTIVE
We compared estimates of UPF consumption derived from 24-hour dietary recall (24HDR) and food frequency questionnaire (FFQ) data in National Health and Nutrition Examination Survey (NHANES) 2003-2006 and assessed agreement between these methods in classifying UPF consumption.
METHODS
This study included 10,937 participants in the NHANES 2003-2004 and 2005-2006 cycles with FFQ and 24HDR data. UPFs were classified according to the Nova classification system. Spearman's rank correlation coefficients and weighted kappa statistics were used to assess agreement between 24HDR and FFQ in estimating UPF intake. The Wilcoxon signed-rank test was used to compare mean UPF intake percentages reported by the two methods. FFQ weights were used to account for sampling.
RESULTS
UPFs accounted for 34.7% of energy using the FFQ and 52.2% of energy intake using the 24HDR, with a mean difference of 18.9% of energy (p < 0.001). This difference was observed across age, sex, and race subgroups. There was a weak positive correlation (Spearman's r: 0.20, p<0.001) and slight agreement (weighted kappa: 0.05) in classifying energy intake from UPF using the two methods. The difference in estimating UPF intake by 24HDR and FFQ was substantially smaller after reclassifying food items on FFQ with ambiguous classification level.
CONCLUSIONS
Differences in UPF intake estimates were consistently observed between FFQ and 24HDR in the overall U.S. population and across demographic subgroups. Nova classification decisions for foods with ambiguous processing levels may contribute to differences between methods, highlighting the need for standardized approaches to classifying foods according to processing level. These findings highlight the need to further the development of dietary assessment instruments designed to capture UPF consumption.
Keyi Jiao, Valerie K. Sullivan, Jiaqi Yang et al.· Journal of NutriLife· 0 citations