Diet during adolescence can have lasting effects on nutritional status, health, and development. We hypothesized that dietary patterns with low-quality nutrition are associated with overweightness. We collected data for 882 Chilean adolescents from the Growth and Obesity Cohort Study (mean age: 12 years). Dietary intake was assessed through 24-h recalls and weight status data were obtained during clinical visits. Dietary patterns were obtained through exploratory factor analysis. Multiple logistic regression models were used to examine cross-sectional associations between dietary patterns and overweight (BMI z-score ≥ 1 SD). Four dietary patterns were identified: “Breakfast/Light dinner”, “Natural foods”, “Western”, and “Snacking”. “Breakfast/Light dinner”, “Western”, and “Snacking” patterns provided higher energy and excess nutrients (sodium, saturated fat, and added sugar). Moreover, adolescents with higher adherence to “Western” or “ Snacking” patterns (third tertile) had higher odds of being classified as overweight (OR = 1.67; 95% CI: 1.103–2.522 and OR = 1.86; 95% CI: 1.235–2.792, respectively) than those with lower adherence (first tertile). “Natural foods” pattern was also associated with overweightness (OR = 1.83; 95% CI: 1.219–2.754). These dietary patterns were associated with overconsumption of nutrients of public health concern. Three of the four main dietary patterns were associated with overweightness. These results highlight the need of prioritizing adolescents on obesity prevention strategies.
Objective: To quantify the energy, nutrients-to-limit and total gram amount consumed and identify their top food sources consumed by Latin Americans.Design: Data from the Latin American Study of Nutrition and Health (ELANS).Setting: ELANS is a cross-sectional study representative of eight Latin American countries: Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, Peru and Venezuela.Participants: Two 24-h dietary recalls on non-consecutive days were used to estimate usual dietary intake of 9218 participants with ages between 15–65 years. ‘What We Eat in America’ food classification system developed by United States Department of Agriculture was adapted and used to classify all food items consumed by the ELANS population. Food sources of energy, added sugars, SFA, Na and total gram amount consumed were identified and ranked based on percentage of contribution to intake of total amount.Results: Three-highest ranked food categories of total energy consumed were: rice (10·3%), yeast breads (6·9%), and turnovers and other grain-based items (6·8 %). Highest ranked food sources of total gram amount consumed were fruit drinks (9·6%), other 100% juice (9·3%) and rice (8·3%). Three highest ranked sources for added sugars were other 100% juice (24·1 %), fruit drinks (16·5%), and sugar and honey (12·4%). SFA ranked foods were turnovers and other grain-based (12·6 %), cheese (11·9%), and pizza (10·3%). Three top sources of Na were rice (13·9%), soups (9·1 %) and rice mixed dishes (7·3 %).Conclusion: Identification of top sources of energy and nutrients-to-limit among Latin Americans is critical for designing strategies to help them meet nutrient recommendations within energy needs.
Deep networks can facilitate the monitoring of a balanced diet to help prevent various health problems related to eating disorders. Large, diverse, and clean data are essential for learning these types of algorithms. Although data can be collected automatically, the data cleaning process is time-consuming. This study aims to provide the model with the ability to learn even when the data are not completely clean. For this purpose, we extend the Attentive Feature MixUp method to enable its learning on noisy multi-label food data. The extension was based on the hypothesis that during the MixUp phase, when a pair of images are mixed, the resulting soft labels should be different for each ingredient, being larger for ingredients that are mixed with the background because they are better distinguished than when they are mixed with other ingredients. Furthermore, to address data perturbation, the incorporation of the Laplace approximation as a post-hoc method was analyzed. The evaluation of the proposed method was performed on two food datasets, where a notable performance improvement was obtained in terms of Jaccard index and score, which validated the hypothesis raised. With the proposed MixUp, our method reduces the memorization of noisy multi-labels, thereby improving its performance.
30 August 2021 in European Journal of Clinical Nutrition
Background: Misreporting of energy intake (EI) in self-reported dietary assessment is inevitable, and even less is known about which food items are misreported by low-middle income adolescents. We evaluated the prevalence of misreporting of energy intake and its relationship with nutrients and food intake. Methods: We analyzed 24 h dietary recalls collected from 576 adolescents (52.08% boys) from southeastern Santiago. Anthropometrics measurements and information about sociodemographic characteristics were obtained during clinical visits. The method proposed by McCrory et al. was used to identify under-reporters (UnRs), over-reporters (OvRs), or plausible reporters (PRs). Food items were collapsed into 28 categories and every food item was expressed as a percentage of total EI. Logistic regression models were performed to investigate the factors associated with misreporting, and a two-part model was used to estimate the difference in the percentage of EI between UnRs versus PRs, and OvRs versus PRs in each food item. Results: Half of the participants were classified as UnRs and 9% were OvRs. UnR was higher among boys (62%) and adolescents with overweight and obesity (72%). OvR was higher among adolescents with normal weight. UnRs had a lower intake of energy from cookies/cake, chocolate/confectionery, and a higher intake of vegetables and eggs than PRs. OvRs had a higher intake of cookies/cake, chocolate/confectionery, and a lower intake of fruit, white milk, and yogurt than PRs. Conclusions: A high frequency of UnR among boys and participants with excess weight was found in this study. Healthy and unhealthy foods are reported differently between UnRs and OvRs of energy intake, indicating that bias is specific for some food items that adolescents commonly eat.
Advancements in artificial intelligence, particularly in computer vision, have driven the research and development of visual food analysis systems focused primarily on enhancing people’s well-being. Food analysis can be performed at various levels of granularity, with food segmentation being a major component of numerous real-world applications. Deep learning-based methodologies have demonstrated promising results in food segmentation; however, many of these approaches demand high computational resources, making them impractical for low-performance devices. In this research, a novel, lightweight, deep learning-based method for semantic food segmentation is proposed. To achieve this, the state-of-the-art DeepLabv3+ model was adapted by optimizing the backbone with the lightweight network EfficientNet-B1, replacing the Atrous Spatial Pyramid Pooling (ASPP) in the neck with Cascade Waterfall ASPP (CWASPP), and refining the encoder output using the squeeze-and-excitation attention mechanism. To validate the method, four publicly available food datasets were selected. Additionally, a new food segmentation dataset consisting of self-acquired food images was introduced and included in the validation. The results demonstrate that high performance can be achieved at a significantly lower cost. The proposed method yields results that are either better than or comparable to those of state-of-the-art techniques while requiring significantly lower computational costs. In conclusion, this research demonstrates the potential of deep learning to perform food image segmentation on low-performance stand-alone devices, paving the way for more efficient, cost-effective, and scalable food analysis applications.