Food classification serves as the basic step of image-based dietary assessment to predict the types of foods in each input image. However, foods in real-world scenarios are typically long-tail ...distributed, where a small number of food types are consumed more frequently than others, which causes a severe class imbalance issue and hinders the overall performance. In addition, none of the existing long-tailed classification methods focus on food data, which can be more challenging due to the inter-class similarity and intra-class diversity between food images. In this work, two new benchmark datasets for long-tailed food classification are introduced, including Food101-LT and VFN-LT, where the number of samples in VFN-LT exhibits real-world long-tailed food distribution. Then, a novel two-phase framework is proposed to address the problem of class imbalance by (1) undersampling the head classes to remove redundant samples along with maintaining the learned information through knowledge distillation and (2) oversampling the tail classes by performing visually aware data augmentation. By comparing our method with existing state-of-the-art long-tailed classification methods, we show the effectiveness of the proposed framework, which obtains the best performance on both Food101-LT and VFN-LT datasets. The results demonstrate the potential to apply the proposed method to related real-life applications.
New imaging technologies to identify food can reduce the reporting burden of participants but heavily rely on the quality of the food image databases to which they are linked to accurately identify ...food images. The objective of this study was to develop methods to create a food image database based on the most commonly consumed U.S. foods and those contributing the most to energy. The objective included using a systematic classification structure for foods based on the standardized United States Department of Agriculture (USDA) What We Eat in America (WWEIA) food classification system that can ultimately be used to link food images to a nutrition composition database, the USDA Food and Nutrient Database for Dietary Studies (FNDDS). The food image database was built using images mined from the web that were fitted with bounding boxes, identified, annotated, and then organized according to classifications aligning with USDA WWEIA. The images were classified by food category and subcategory and then assigned a corresponding USDA food code within the USDA's FNDDS in order to systematically organize the food images and facilitate a linkage to nutrient composition. The resulting food image database can be used in food identification and dietary assessment.
The objective was to determine the most frequently consumed food items, food subcategories, and food categories, and those that contributed most to total energy intake for the group of U.S. adults ...reporting taking insulin, those with type 2 diabetes (T2D) not taking insulin, and those without diabetes. Laboratory tests and questionnaires of the National Health and Nutrition Examination Survey 2009–2016 classified 774 participants reporting taking insulin, 2758 participants reporting T2D not taking insulin, and 17,796 participants without diabetes. Raw and weighted frequency and energy contributions of each food item, food subcategory, and food category were calculated and ranked. Comparisons among groups by broad food category used the Rao–Scott modified chi-square test. Soft drinks ranked as the 8th and 6th most consumed food subcategory of participants with T2D not taking insulin and those without diabetes, and contributed 5th and 2nd most to energy, respectively. The group reporting taking insulin is likely to consume more protein foods and less soft drink compared to the other two groups. Lists of the most frequently reported foods and foods contributing most to energy may be helpful for nutrition education, prescribing diets, and digital-based dietary assessment for the group reporting taking insulin.
Data-driven temporal dietary patterning (TDP) methods were previously developed. The objectives were to create data-driven temporal dietary patterns and assess concurrent validity of energy and time ...cut-offs describing the data-driven TDPs by determining their relationships to BMI and waist circumference (WC). The first day 24-h dietary recall timing and amounts of energy for 17,915 U.S. adults of the National Health and Nutrition Examination Survey 2007−2016 were used to create clusters representing four TDPs using dynamic time warping and the kernel k-means clustering algorithm. Energy and time cut-offs were extracted from visualization of the data-derived TDPs and then applied to the data to find cut-off-derived TDPs. The strength of TDP relationships with BMI and WC were assessed using adjusted multivariate regression and compared. Both methods showed a cluster, representing a TDP with proportionally equivalent average energy consumed during three eating events/day, associated with significantly lower BMI and WC compared to the other three clusters that had one energy intake peak/day at 13:00, 18:00, and 19:00 (all p < 0.0001). Participant clusters of the methods were highly overlapped (>83%) and showed similar relationships with obesity. Data-driven TDP was validated using descriptive cut-offs and hold promise for obesity interventions and translation to dietary guidance.
Energy and dietary quality are known to differ between weekdays and weekends. Data-driven approaches that incorporate time, amount, and duration of dietary intake have previously been used to ...partition participants’ daily weekday dietary intake time series into clusters representing weekday temporal dietary patterns (TDPs) linked to health indicators in United States adults. Yet, neither the relationship of weekend day TDPs to health indicators nor how the TDP membership may change from weekday to weekend is known.
This study was conducted to determine the association between TDPs on weekdays and weekend days and health indicators diet quality, waist circumference (WC), body mass index (BMI), and obesity and their overlap among participants.
A weekday and weekend day 24-hour dietary recall of 9494 nonpregnant United States adults aged 20–65 years from the cross-sectional National Health and Nutrition Examination Survey 2007–2018 was used to determine the timing and amount of energy intake. Modified dynamic time warping and kernel k-means algorithm clustered participants into 4 TDPs on weekdays and weekend days. Multivariate regression models determined the associations between TDPs and health indicators, controlling for potential confounders and adjusting for the survey design and multiple comparisons. The percentages of overlap in cluster membership between TDPs on weekdays and weekend days were also determined.
United States adults with a TDP of evenly spaced, energy-balanced eating occasions, representing the TDP of more than one-third of all adults on weekdays and weekends, had significantly higher diet quality, lower BMI, WC, and odds of obesity when compared to those with other TDPs. Membership of most United States adults to TDPs varied from weekdays to weekends.
Both weekday and weekend TDPs were significantly associated with health indicators. TDP membership of most United States adults was not consistent on weekdays and weekends.
Few attempts have been made to incorporate multiple aspects of physical activity (PA) to classify patterns linked with health. Temporal PA patterns integrating time and activity counts were created ...to determine their association with health status. Accelerometry data from the National Health and Nutrition Examination Survey 2003–2006 was used to pattern PA counts and time of activity from 1999 adults with one weekday of activity. Dynamic time warping and kernel k-means clustering partitioned 4 participant clusters representing temporal PA patterns. Multivariate regression models determined associations between clusters and health status indicators and obesity, type 2 diabetes, and metabolic syndrome. Cluster 1 with a temporal PA pattern of the lowest activity counts reaching 4.8e4 cph from 6:00–23:00 was associated with higher body mass index (BMI) (β = 2.5 ± 0.6 kg/m2, 95% CI: 1.0, 4.1), higher waist circumference (WC) (β = 6.4 ± 1.3 cm, 95% CI: 2.8, 10.0), and higher odds of obesity (OR: 2.4; 95% CI: 1.3, 4.4) compared with Cluster 3 with activity counts reaching 9.6e4–1.2e5 cph between 16:00–21:00. Cluster 1 was also associated with higher BMI (β = 1.5 ± 0.5 kg/m2, 95% CI: 0.1, 2.8) and WC (β = 3.6 ± 1.3 cm, 95% CI: 0.1, 7.0) compared to Cluster 4 with activity counts reaching 9.6e4 cph between 8:00–11:00. A Temporal PA pattern with the lowest PA counts had significantly higher mean BMI and WC compared to temporal PA patterns of higher activity counts performed early (8:00–11:00) or late (16:00–21:00) throughout the day. Temporal PA patterns appear to meaningfully link to health status.
•Temporal patterns created by a data-driven approach link meaningfully with health.•Dynamic time warping and kernel-k means for time-based physical activity patterns.•Joining time and physical activity for multidimensional patterns linked to health.•Temporal patterns of higher physical activity counts related to health status.•Temporal activity patterns are associated with BMI and waist circumference.
The integration of time with dietary patterns throughout a day, or temporal dietary patterns (TDPs), have been linked with dietary quality but relations to health are unknown.
The association between ...TDPs and selected health status indicators and obesity, type 2 diabetes (T2D), and metabolic syndrome (MetS) was determined.
The first-day 24-h dietary recall from 1627 nonpregnant US adult participants aged 20–65 y from the NHANES 2003–2006 was used to determine timing, amount of energy intake, and sequence of eating occasions (EOs). Modified dynamic time warping (MDTW) and kernel k-means algorithm clustered participants into 4 groups representing distinct TDPs. Multivariate regression models determined associations between TDPs and health status, controlling for potential confounders, and adjusting for the survey design and multiple comparisons (P <0.05/6).
A cluster representing a TDP with evenly spaced, energy balanced EOs reaching ≤1200 kcal between 06:00 to 10:00, 12:00 to 15:00, and 18:00 to 22:00, had statistically significant and clinically meaningful lower mean BMI (P <0.0001), waist circumference (WC) (P <0.0001), and 75% lower odds of obesity compared with 3 other clusters representing patterns with much higher peaks of energy: 1000–2400 kcal between 15:00 and 18:00 (OR: 5.3; 95% CI: 2.8, 10.1), 800–2400 kcal between 11:00 and 15:00 (OR: 4.4; 95% CI: 2.5, 7.9), and 1000–2600 kcal between 18:00 and 23:00 (OR: 6.7; 95% CI: 3.9, 11.6).
Individuals with a TDP characterized by evenly spaced, energy balanced EOs had significantly lower mean BMI, WC, and odds of obesity compared with the other patterns with higher energy intake peaks at different times throughout the day, providing evidence that incorporating time with other aspects of a dietary pattern may be important to health status.
Daily temporal patterns of energy intake (temporal dietary patterns TDPs) and physical activity (temporal physical activity patterns TPAPs) have been independently and jointly (temporal dietary and ...physical activity patterns TDPAPs) associated with health and disease status indicators.
The aim of this study was to compare the number and strength of association between clusters of daily TDPs, TPAPs, and TDPAPs and multiple health and disease status indicators.
This cross-sectional study used 1 reliable weekday dietary recall and 1 random weekday of accelerometer data to partition to create clusters of participants representing the 3 temporal patterns. Four clusters were created via kernel-k means clustering algorithm of the same constrained dynamic time warping distance computed over the time series for each temporal pattern.
From the National Health and Nutrition Examination Survey (2003-2006), 1,836 US adults aged 20 through 65 years who were not pregnant and had valid diet, physical activity, sociodemographic, anthropometric, questionnaire, and health and disease status indicator data were included.
Health status indicators used as outcome measures were body mass index, waist circumference, fasting plasma glucose, hemoglobin A1c, triglycerides, high-density lipoprotein cholesterol, total cholesterol, and systolic and diastolic blood pressure; disease status indicators included obesity, type 2 diabetes mellitus, and metabolic syndrome.
Multivariate regression models determined associations between the clusters representing each pattern and health and disease status indicators, controlling for confounders and adjusting for multiple comparisons. The number of significant differences among clusters and adjusted R
and Akaike information criterion compared the strength of associations between clusters of patterns and continuous and categorical health and disease status indicators.
TDPAPs showed 21 significant associations with health and disease status indicators, including body mass index, waist circumference, obesity, and type 2 diabetes; TDPs showed 19 significant associations; and TPAPs showed 8 significant associations.
TDPAPs and TDPs had stronger and more numerous associations with health and disease status indicators compared with TPAPs. Patterns representing the integration of daily dietary habits hold promise for early detection of obesity.
Diet and physical activity (PA) are independent risk factors for obesity and chronic diseases including type 2 diabetes mellitus (T2DM) and metabolic syndrome (MetS). The temporal sequence of these ...exposures may be used to create patterns with relations to health status indicators.
The objectives were to create clusters of joint temporal dietary and PA patterns (JTDPAPs) and to determine their association with health status indicators including BMI, waist circumference (WC), fasting plasma glucose, glycated hemoglobin, triglycerides, HDL cholesterol, total cholesterol, blood pressure, and disease status including obesity, T2DM, and MetS in US adults.
A 24-h dietary recall and random day of accelerometer data of 1836 participants from the cross-sectional NHANES 2003–2006 data were used to create JTDPAP clusters by constrained dynamic time warping, coupled with a kernel k-means clustering algorithm. Multivariate regression models determined associations between the 4 JTDPAP clusters and health and disease status indicators, controlling for potential confounders and adjusting for multiple comparisons.
A JTDPAP cluster with proportionally equivalent energy consumed at 2 main eating occasions reaching ≤1600 and ≤2200 kcal from 11:00 to 13:00 and from 17:00 to 20:00, respectively, and the highest PA counts among 4 clusters from 08:00 to 20:00, was associated with significantly lower BMI (P < 0.0001), WC (P = 0.0001), total cholesterol (P = 0.02), and odds of obesity (OR: 0.2; 95% CI: 0.1, 0.5) than a JTDPAP cluster with proportionally equivalent energy consumed reaching ≤1600 and ≤1800 kcal from 11:00 to 14:00 and from 17:00 to 21:00, respectively, and high PA counts from 09:00 to 12:00.
The joint temporally patterned sequence of diet and PA can be used to cluster individuals with meaningful associations to BMI, WC, total cholesterol, and obesity. Temporal patterns hold promise for future development of lifestyle patterns that integrate additional temporal and contextual activities.
Epidemiologic evidence supports a positive association between ultraprocessed food (UPF) consumption and body mass index. This has led to recommendations to avoid UPFs despite very limited evidence ...establishing causality. Many mechanisms have been proposed, and this review critically aimed to evaluate selected possibilities for specificity, clarity, and consistency related to food choice (i.e., low cost, shelf-life, food packaging, hyperpalatability, and stimulation of hunger/suppression of fullness); food composition (i.e., macronutrients, food texture, added sugar, fat and salt, energy density, low-calorie sweeteners, and additives); and digestive processes (i.e., oral processing/eating rate, gastric emptying time, gastrointestinal transit time, and microbiome). For some purported mechanisms (e.g., fiber content, texture, gastric emptying, and intestinal transit time), data directly contrasting the effects of UPF and non-UPF intake on the indices of appetite, food intake, and adiposity are available and do not support a unique contribution of UPFs. In other instances, data are not available (e.g., microbiome and food additives) or are insufficient (e.g., packaging, food cost, shelf-life, macronutrient intake, and appetite stimulation) to judge the benefits versus the risks of UPF avoidance. There are yet other evoked mechanisms in which the preponderance of evidence indicates ingredients in UPFs actually moderate body weight (e.g., low-calorie sweetener use for weight management; beverage consumption as it dilutes energy density; and higher fat content because it reduces glycemic responses). Because avoidance of UPFs holds potential adverse effects (e.g., reduced diet quality, increased risk of food poisoning, and food wastage), it is imprudent to make recommendations regarding their role in diets before causality and plausible mechanisms have been verified.
PDF
