Article

The obesogenity of restaurant food: Mapping the nutritional foodscape of Franklin County, Ohio using food review images

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract and Figures

Geographical research on the community food environment emphasizes the spatial dimension of food access, where food quality and nutrition are overlooked. While the nutritional quality of food stores can be evaluated by audit tools, such as the Nutrition Environment Measurement Survey series, these tools cannot be readily applied on a large scale. To aid in the nutrition assessment of the community food environment, this article proposes an innovative image recognition approach that identifies the nutrition information of restaurant food by crowdsourcing food images and retrieving their nutrition information in a case study of Franklin County, Ohio. As these food images are derived from food review websites, they represent consumers' actual dietary quality as a more direct, quantifiable measure of the community food environment. The derived nutrition information is applied for further analysis to characterize neighborhood obesogenity, referring to as factors in the built environments that facilitate the development of obesity. The proposed crowdsourcing-image recognition approach enriches the understanding of health inequities from a nutrition science perspective, which is beyond the tradition of food access in the spatial dimension only.
Content may be subject to copyright.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Deep learning models can recognize the food item in an image and derive their nutrition information, including calories, macronutrients (carbohydrates, fats, and proteins), and micronutrients (vitamins and minerals). This technology has yet to be implemented for the nutrition assessment of restaurant food. In this paper, we crowdsource 15,908 food images of 470 restaurants in the Greater Hartford region on Tripadvisor and Google Place. These food images are loaded into a proprietary deep learning model (Calorie Mama) for nutrition assessment. We employ manual coding to validate the model accuracy based on the Food and Nutrient Database for Dietary Studies. The derived nutrition information is visualized at both the restaurant level and the census tract level. The deep learning model achieves 75.1% accuracy when compared with manual coding. It has more accurate labels for ethnic foods but cannot identify portion sizes, certain food items (e.g., specialty burgers and salads), and multiple food items in an image. The restaurant nutrition (RN) index is further proposed based on the derived nutrition information. By identifying the nutrition information of restaurant food through crowdsourced food images and a deep learning model, the study provides a pilot approach for large-scale nutrition assessment of the community food environment.
Article
Full-text available
Menu calorie labeling is now required nationwide for chain restaurants in the United States; however, a number of studies have found that calorie labeling does not reduce average calories ordered. This research examines how different food value orientations are associated with divergent consumer responses to restaurants providing calorie information on menus and menu boards. Results from two pilot studies and two experiments, including a restaurant field experiment, indicate that calorie labeling is effective in decreasing the number of calories ordered by health value–oriented consumers. However, for quantity value and taste value–oriented consumers, menu calorie labeling may result in an increase in calories ordered. These influences counterbalance one another, leading to an overall nonsignificant effect of calorie labeling on calories ordered in restaurant settings. These findings offer a compelling explanation for the many studies showing nonsignificant effects of menu calorie labeling and inform ongoing policy debates regarding chain restaurants nationally implementing menu calorie labeling. The conceptual contributions and implications of these findings for public policy and consumer well-being are discussed.
Article
Full-text available
Article
Full-text available
This paper investigates the effect of food environments, characterized as food swamps, on adult obesity rates. Food swamps have been described as areas with a high-density of establishments selling high-calorie fast food and junk food, relative to healthier food options. This study examines multiple ways of categorizing food environments as food swamps and food deserts, including alternate versions of the Retail Food Environment Index. We merged food outlet, sociodemographic and obesity data from the United States Department of Agriculture (USDA) Food Environment Atlas, the American Community Survey, and a commercial street reference dataset. We employed an instrumental variables (IV) strategy to correct for the endogeneity of food environments (i.e., that individuals self-select into neighborhoods and may consider food availability in their decision). Our results suggest that the presence of a food swamp is a stronger predictor of obesity rates than the absence of full-service grocery stores. We found, even after controlling for food desert effects, food swamps have a positive, statistically significant effect on adult obesity rates. All three food swamp measures indicated the same positive association, but reflected different magnitudes of the food swamp effect on rates of adult obesity (p values ranged from 0.00 to 0.16). Our adjustment for reverse causality, using an IV approach, revealed a stronger effect of food swamps than would have been obtained by naïve ordinary least squares (OLS) estimates. The food swamp effect was stronger in counties with greater income inequality (p < 0.05) and where residents are less mobile (p < 0.01). Based on these findings, local government policies such as zoning laws simultaneously restricting access to unhealthy food outlets and incentivizing healthy food retailers to locate in underserved neighborhoods warrant consideration as strategies to increase health equity.
Article
Full-text available
Most research on urban food deserts has employed spatial measures of accessibility, recording distances to various food stores from place of residence. Despite the popularity of this approach, empirical support for its prediction of dietary and health outcomes has been inconsistent. One reason might be the ways in which food deserts frame food access as fundamentally an issue of food supply. This article suggests a complementary approach that examines how store characteristics, neighborhood context, and individual mobility interact to shape food provisioning practices. I recruited thirty-eight participants living in two low-income neighborhoods of Minneapolis, Minnesota, tracking their daily mobility and the food sources they used over a five-day study period. Follow-up interviews gathered more information on the food stores used by participants. Project results show that participants were highly mobile in their food shopping, visiting 153 different locations on 217 different shopping trips at an average distance of 3.4 km from home. Reported store quality was closely tied to neighborhoods' economic and racial composition, and in several cases, food purchasing and consumption occurred en route to other destinations. Future research on urban food access could benefit by studying how food access is intertwined with broader livability issues such as housing and transportation.
Article
Full-text available
The Digital Photography of Foods Method accurately estimates the food intake of adults and children in cafeterias. When using this method, images of food selection and leftovers are quickly captured in the cafeteria. These images are later compared with images of 'standard' portions of food using computer software. The amount of food selected and discarded is estimated based upon this comparison, and the application automatically calculates energy and nutrient intake. In the present review, we describe this method, as well as a related method called the Remote Food Photography Method (RFPM), which relies on smartphones to estimate food intake in near real-time in free-living conditions. When using the RFPM, participants capture images of food selection and leftovers using a smartphone and these images are wirelessly transmitted in near real-time to a server for analysis. Because data are transferred and analysed in near real-time, the RFPM provides a platform for participants to quickly receive feedback about their food intake behaviour and to receive dietary recommendations for achieving weight loss and health promotion goals. The reliability and validity of measuring food intake with the RFPM in adults and children is also reviewed. In sum, the body of research reviewed demonstrates that digital imaging accurately estimates food intake in many environments and it has many advantages over other methods, including reduced participant burden, elimination of the need for participants to estimate portion size, and the incorporation of computer automation to improve the accuracy, efficiency and cost-effectiveness of the method.
Article
Full-text available
Objective: The present study aimed to (i) describe the availability of nutrition information in major chain restaurants, (ii) document the energy and nutrient levels of menu items, (iii) evaluate relationships with restaurant characteristics, menu labelling and trans fat laws, and nutrition information accessibility, and (iv) compare energy and nutrient levels against industry-sponsored and government-issued nutrition criteria. Design: Descriptive statistics and multivariate regression analysis of the energy, total fat, saturated fat, trans fat, sodium, carbohydrate and protein levels of 29 531 regular and 1392 children's menu items [corrected]. Setting: Energy and nutrition information provided on restaurant websites or upon request, and secondary databases on restaurant characteristics. Subjects: The top 400 US chain restaurants by sales, based on the 2009 list of the Restaurants & Institutions magazine. Results: Complete nutrition information was reported for 245 (61 %) restaurants. Appetizers had more energy, fat and sodium than all other item types. Children's menu specialty beverages had more fat, saturated fat and carbohydrates than comparable regular menu beverages. The majority of main entrées fell below one-third of the US Department of Agriculture's estimated daily energy needs, but as few as 3 % were also within limits for sodium, fat and saturated fat. Main entrées had significantly more energy, fat and saturated fat in family-style restaurants than in fast-food restaurants. Restaurants that made nutrition information easily accessible on websites had significantly lower energy, fat and sodium contents across menu offerings than those providing information only upon request. Conclusions: The paper provides a comprehensive view of chain restaurant menu nutrition prior to nationwide labelling laws. It offers baseline data to evaluate how restaurants respond after laws are implemented.
Article
Full-text available
Despite pervasive exurban development in the United States (US) over the last several decades, a lack of relatively precise data has hindered basic research, including classification of the types of emerging exurban settlement patterns. But because exurbia transcends the traditional dichotomy of urban versus rural and metropolitan versus nonmetropolitan, standard census data are of limited value in analyzing these regions. To address this gap, we made use of a spatially explicit population database to indentify and examine the configuration of exurban settlement. We developed a multidimensional spatial patch index, which captures the variation exurban settlement across the landscape, along the lines of shape, size and contiguity typology of exurban patches. This index reveals substantial heterogeneity in fine-scale pattern not reflected by aggregate measures of exurban settlement density.Using this index, we then explored to what extent metropolitan characteristics commonly believed to influence urban decentralization also are associated with the configuration of decentralization. We find that while large, contiguous clumps of exurban settlement are correlated with these metropolitan characteristics, other exurban pattern types (such as isolated or linear patches) are not. Therefore, it appears that the factors that contribute to more urban decentralization may also lead to greater contiguity and concentration of exurban settlement. Additionally, we find that the processes that lead to isolated and dispersed exurban settlement may be different from those that foster contiguous, concentrated exurban settlement.
Article
Full-text available
Why is America a land of low-calorie food claims yet high-calorie food intake? Four studies show that people are more likely to underestimate the caloric content of main dishes and to choose higher-calorie side dishes, drinks, or desserts when fast-food restaurants claim to be healthy ( e. g., Subway) compared to when they do not ( e. g., McDonald's). We also find that the effect of these health halos can be eliminated by simply asking people to consider whether the opposite of such health claims may be true. These studies help explain why the success of fast-food restaurants serving lower-calorie foods has not led to the expected reduction in total calorie intake and in obesity rates. They also suggest innovative strategies for consumers, marketers, and policy makers searching for ways to fight obesity.
Article
Full-text available
To examine the association between the increase in body mass index (BMI) and socioeconomic factors (eg, income level, % below poverty line, unemployment rates and persons receiving food stamps) in Mississippi, Alabama, Louisiana, Tennessee and Colorado. Data from Behavioral Risk Factor Surveillance System, United States Department of Agriculture and the United States Department of Labor/Bureau of Labor were obtained and analyzed for the years 1995-2008. Results from this study showed a strong association between obesity and the tested variables (R2 = .767). Factors more closely related with obesity were: income below poverty level; receipt of food stamps; unemployment; and general income level. The coefficient of determination for these variables were 0.438, 0.427, 0.103 and 0.018, respectively. The highest rate of obesity was found in Mississippi (26.5% +/- 4.13%) followed by Alabama (25.18% +/- 4.41%), while Colorado had the lowest rate of obesity (15.4% +/- 2.63%). By ethnicity, African Americans had the highest rate of obesity (32.64 +/- 5.99%). We found a significant effect of consumption of low-quality food, due to economic factors, on increased BMI. Besides physical activity, the quality and the quantity of food are important factors that contribute to obesity rates.
Article
Full-text available
Through a literature review, we investigated the geographic information systems (GIS) methods used to define the food environment and the types of spatial measurements they generate. Review study. Searches were conducted in health science databases, including Medline/Pubmed, PsycINFO, Francis and GeoBase. We included studies using GIS-based measures of the food environment published up to 1 June 2008. Twenty-nine papers were included. Two different spatial approaches were identified. The density approach quantifies the availability of food outlets using the buffer method, kernel density estimation or spatial clustering. The proximity approach assesses the distance to food outlets by measuring distances or travel times. GIS network analysis tools enable the modelling of travel time between referent addresses (home) and food outlets for a given transportation network and mode, and the assumption of travel routing behaviours. Numerous studies combined both approaches to compare food outlet spatial accessibility between different types of neighbourhoods or to investigate relationships between characteristics of the food environment and individual food behaviour. GIS methods provide new approaches for assessing the food environment by modelling spatial accessibility to food outlets. On the basis of the available literature, it appears that only some GIS methods have been used, while other GIS methods combining availability and proximity, such as spatial interaction models, have not yet been applied to this field. Future research would also benefit from a combination of GIS methods with survey approaches to describe both spatial and social food outlet accessibility as important determinants of individual food behaviours.
Article
Full-text available
Obesity and inadequate physical activity are major risk factors for many diseases. The built environment plays an important role in influencing participation in physical activity. We aimed to determine whether urban sprawl in Sydney, Australia is associated with overweight/obesity and levels of physical activity. We used a cross-sectional multilevel study design to relate urban sprawl (based on population density) measured at an area level to overweight/obesity and levels of physical activity measured at an individual level whilst controlling for individual and area level covariates in metropolitan Sydney. Individual level data were obtained from the 2002 and 2003 New South Wales Population Health Survey. We had information on 7,290 respondents. The mean population density was 2,168 persons per square kilometer (standard deviation=1,741, range=218-7,045). After controlling for individual and area level covariates, for an inter-quartile increase in sprawl, the odds of being overweight was 1.26 (95% CI=1.10-1.44), the odds of being obese was 1.47 (95% CI=1.24-1.75), the odds of inadequate physical activity was 1.38 (95% CI=1.21-1.57), and the odds of not spending any time walking during the past week was 1.58 (95% CI=1.28-1.93). Living in more sprawling suburbs increases the risk of overweight/obesity and inadequate physical activity despite the relatively low levels of urban sprawl in metropolitan Sydney. Modifications to the urban environment to increase physical activity may be worthwhile.
Article
Full-text available
Dependence on motorized forms of transportation may contribute to the worldwide obesity epidemic. Shifts in transportation patterns occurring in China provide an ideal opportunity to study the association between vehicle ownership and obesity. Our objective was to determine whether motorized forms of transportation promote obesity. A multistage random-cluster sampling process was used to select households from eight provinces in China. Data were included on household vehicle ownership and individual anthropometric and sociodemographic status. Cross-sectional data (1997) from 4741 Chinese adults aged 20 to 55 years were used to explore the association between vehicle ownership and obesity. Cohort data (1989 to 1997) from 2485 adults aged 20 to 45 years in 1989 (59% follow-up) were used to measure the impact of vehicle acquisition on the odds of becoming obese. Our main outcome measure was current obesity status and the odds of becoming obese over an 8-year period. In 1997, 84% of adults did not own motorized transportation. However, the odds of being obese were 80% higher (p < 0.05) for men and women in households who owned a motorized vehicle compared with those who did not own a vehicle. Fourteen percent of households acquired a motorized vehicle between 1989 and 1997. Compared with those whose vehicle ownership did not change, men who acquired a vehicle experienced a 1.8-kg greater weight gain (p < 0.05) and had 2 to 1 odds of becoming obese. Encouraging active forms of transportation may be one way to protect against obesity.
Article
Full-text available
The prevalence of overweight among children in the United States increased between 1976-1980 and 1988-1994, but estimates for the current decade are unknown. To determine the prevalence of overweight in US children using the most recent national data with measured weights and heights and to examine trends in overweight prevalence. Survey of 4722 children from birth through 19 years of age with weight and height measurements obtained in 1999-2000 as part of the National Health and Nutrition Examination Survey (NHANES), a cross-sectional, stratified, multistage probability sample of the US population. Prevalence of overweight among US children by sex, age group, and race/ethnicity. Overweight among those aged 2 through 19 years was defined as at or above the 95th percentile of the sex-specific body mass index (BMI) for age growth charts. The prevalence of overweight was 15.5% among 12- through 19-year-olds, 15.3% among 6- through 11-year-olds, and 10.4% among 2- through 5-year-olds, compared with 10.5%, 11.3%, and 7.2%, respectively, in 1988-1994 (NHANES III). The prevalence of overweight among non-Hispanic black and Mexican-American adolescents increased more than 10 percentage points between 1988-1994 and 1999-2000. The prevalence of overweight among children in the United States is continuing to increase, especially among Mexican-American and non-Hispanic black adolescents.
Article
Full-text available
Research in transportation, urban design, and planning has examined associations between physical environment variables and individuals' walking and cycling for transport. Constructs, methods, and findings from these fields can be applied by physical activity and health researchers to improve understanding of environmental influences on physical activity. In this review, neighborhood environment characteristics proposed to be relevant to walking/cycling for transport are defined, including population density, connectivity, and land use mix. Neighborhood comparison and correlational studies with nonmotorized transport outcomes are considered, with evidence suggesting that residents from communities with higher density, greater connectivity, and more land use mix report higher rates of walking/cycling for utilitarian purposes than low-density, poorly connected, and single land use neighborhoods. Environmental variables appear to add to variance accounted for beyond sociodemographic predictors of walking/cycling for transport. Implications of the transportation literature for physical activity and related research are outlined. Future research directions are detailed for physical activity research to further examine the impact of neighborhood and other physical environment factors on physical activity and the potential interactive effects of psychosocial and environmental variables. The transportation, urban design, and planning literatures provide a valuable starting point for multidisciplinary research on environmental contributions to physical activity levels in the population.
Article
Full-text available
The authors provide a conceptual model of a healthy nutrition environment, then review the types of measures required to assess various aspects of this environment. Measures fall into priority categories of consumer and community environments.
Article
Full-text available
This is a short report of a 'safari' held in conjunction with the International Congress of Nutrition in September 2005, in Futululu, St. Lucia, South Africa. Participants were several members of the International Union of Nutritional Sciences Task Force on Indigenous Peoples' Food Systems and Nutrition, other interested scientists and members of the Kwa Zulu indigenous community. The paper describes the rationale for and contributions towards understanding what might be successful interventions that would resonate among indigenous communities in many areas of the world. A summary of possible evaluation strategies of such interventions is also given.
Article
Full-text available
The aim of the present study was to report the first reliability and validity tests of the remote food photography method (RFPM), which consists of camera-enabled cell phones with data transfer capability. Participants take and transmit photographs of food selection and plate waste to researchers/clinicians for analysis. Following two pilot studies, adult participants (n 52; BMI 20-35 kg/m2 inclusive) were randomly assigned to the dine-in or take-out group. Energy intake (EI) was measured for 3 d. The dine-in group ate lunch and dinner in the laboratory. The take-out group ate lunch in the laboratory and dinner in free-living conditions (participants received a cooler with pre-weighed food that they returned the following morning). EI was measured with the RFPM and by directly weighing foods. The RFPM was tested in laboratory and free-living conditions. Reliability was tested over 3 d and validity was tested by comparing directly weighed EI to EI estimated with the RFPM using Bland-Altman analysis. The RFPM produced reliable EI estimates over 3 d in laboratory (r 0.62; P < 0.0001) and free-living (r 0.68; P < 0.0001) conditions. Weighed EI correlated highly with EI estimated with the RFPM in laboratory and free-living conditions (r>0.93; P < 0.0001). In two laboratory-based validity tests, the RFPM underestimated EI by - 4.7 % (P = 0.046) and - 5.5 % (P = 0.076). In free-living conditions, the RFPM underestimated EI by - 6.6 % (P = 0.017). Bias did not differ by body weight or age. The RFPM is a promising new method for accurately measuring the EI of free-living individuals. Error associated with the method is small compared with self-report methods.
Article
Applied Geography is a journal devoted to the publication of research which utilizes geographic approaches (human, physical, nature-society and GIScience) to resolve human problems that have a spatial dimension. These problems may be related to the assessment, management and allocation of the world's physical and/or human resources. The underlying rationale of the journal is that only through a clear understanding of the relevant societal, physical, and coupled natural-humans systems can we resolve such problems. Applied Geography is a journal devoted to the publication of research which utilizes geographic approaches (human, physical, nature-society and GIScience) to resolve human problems that have a spatial dimension. These problems may be related to the assessment, management and allocation of the world's physical and/or human resources. The underlying rationale of the journal is that only through a clear understanding of the relevant societal, physical, and coupled natural-humans systems can we resolve such problems. 2
Article
Background: Prior research on the restaurant environment and obesity risk is limited by cross-sectional data and a focus on specific geographic areas. Objective: To measure the impact of changes in chain restaurant calories over time on body mass index (BMI). Design: We used a first-difference model to examine whether changes from 2012 to 2015 in chain restaurant calories per capita were associated with percent changes in BMI. We also examined differences by race and county income, restaurant type, and initial body weight categories. Setting: USA (207 counties across 39 states). Participants: 447,873 adult patients who visited an athenahealth medical provider in 2012 and 2015 where BMI was measured. Main outcomes measured: Percent change in objectively measured BMI from 2012 to 2015. Results: Across all patients, changes in chain restaurant calories per capita were not associated with percent changes in BMI. For Black or Hispanic adults, a 10% increase in exposure to chain restaurant calories per capita was associated with a 0.16 percentage-point increase in BMI (95% CI 0.03, 0.30). This translates into a predicted weight increase of 0.89 pounds (or a 0.53% BMI increase) for an average weight woman at the 90th percentile of increases in the restaurant environment from 2012 to 2015 versus an increase 0.39 pounds (or 0.23% BMI increase) at the 10th percentile. Greater increases in exposure to chain restaurant calories also significantly increased BMI for Black or Hispanic adults receiving healthcare services in lower-income counties (0.26, 95% CI 0.04, 0.49) and with overweight/obesity (0.16, 95% CI 0.04, 0.29). Limitations: Generalizability to non-chain restaurants is unknown and the sample of athenahealth patients is relatively homogenous. Conclusions: Increased exposure to chain restaurant calories per capita was associated with increased weight gain among Black or Hispanic adults.
Article
Introduction: The objective of this study is to investigate the early impact of Canada's first provincewide mandatory menu-labeling legislation on calorie levels in foods offered on chain restaurant menus before, leading up to, and at the point-of-implementation. Methods: Data were obtained from Menu-FLIP, a database of publicly available nutrition data from chain restaurants in Canada. Data were collected in 2010, 2013, 2016, and 2017. Core food products, newly introduced products, discontinued products, and full menus were compared before (2010, 2013, 2016) and at the point-of-implementation (2017) of the policy. Calories and serving sizes of 2,988 unique foods (excluding beverages) from 28 chains were analyzed using mixed models comparing time points while controlling for covariates. Sub-analyses were stratified by restaurant type. Results: Full menu mean calories and serving sizes increased between 2010 and 2017 (>40 calories and 17 grams, respectively). There were no significant changes in mean calories or serving sizes among core menu items (p=0.47). New products in 2013 and 2017 were significantly higher in calories and serving size versus discontinued foods and foods already on the menu. Discontinued foods did not differ in calories or serving size compared with foods already on the menu; however, in 2017, the calories of discontinued foods in sit-down restaurants were higher than foods discontinued in 2013 (p=0.02). Conclusions: The results of this investigation show that at the point-of-implementation of the Ontario Healthy Menu Choices Act supply-side calories in core food products, new products, or full menus evaluated in this study did not significantly change.
Article
The food desert metaphor has been widely used over the past few decades as a way to identify regions as being at risk for having little or no access to healthy food. While the simplicity of the metaphor is attractive, this article argues that its usefulness to researchers interested in understanding the relationship between the geography of healthy food opportunities and dietary behaviours is limited. More nuanced approaches to incorporating geography into food access studies, like including transportation, economic factors, and time use, in addition to considering other dimensions of accessibility, are warranted.
Book
The phenomenon of volunteered geographic information is part of a profound transformation in how geographic data, information, and knowledge are produced and circulated. By situating volunteered geographic information (VGI) in the context of big-data deluge and the data-intensive inquiry, the 20 chapters in this book explore both the theories and applications of crowdsourcing for geographic knowledge production with three sections focusing on 1). VGI, Public Participation, and Citizen Science; 2). Geographic Knowledge Production and Place Inference; and 3). Emerging Applications and New Challenges. This book argues that future progress in VGI research depends in large part on building strong linkages with diverse geographic scholarship. Contributors of this volume situate VGI research in geography’s core concerns with space and place, and offer several ways of addressing persistent challenges of quality assurance in VGI. This book positions VGI as part of a shift toward hybrid epistemologies, and potentially a fourth paradigm of data-intensive inquiry across the sciences. It also considers the implications of VGI and the exaflood for further time-space compression and new forms, degrees of digital inequality, the renewed importance of geography, and the role of crowdsourcing for geographic knowledge production.
Article
This paper discusses measurement of the main dimensions of the urban environment that have been proposed as relevant to explaining geographic variations in obesity and inactivity. It considers urban sprawl, food access and exercise access as latent constructs, defined by sets of observed indicators for areas. In an application to 993 US metropolitan counties, the paper shows how these latent constructs may be incorporated in an ecological (area-scale) model, which recognizes spatial aspects in the patterning of both outcomes and environmental factors. Urban sprawl and area socioeconomic status emerge from regression modelling as leading influences on obesity and inactivity.
Article
In this paper, we report the feature obtained from the Deep Convolutional Neural Network boosts food recognition accuracy greatly by integrating it with conventional hand-crafted image features, Fisher Vectors with HoG and Color patches. In the experiments, we have achieved 72.26% as the top-1 accuracy and 92.00% as the top-5 accuracy for the 100-class food dataset, UEC-FOOD100, which outperforms the best classification accuracy of this dataset reported so far, 59.6%, greatly.
Article
We examined the uncertainty of the contextual influences on food access through an analytic framework of the uncertain geographic context problem (UGCoP). We first examined the compounding effects of two kinds of spatiotemporal uncertainties on people's everyday efforts to procure food and then outlined three key dimensions (food access in real time, temporality of the food environment, and perceived nutrition environment) in which research on food access must improve to better represent the contributing environmental influences that operate at the individual level. Guidelines to address the UGCoP in future food access research are provided to account for the multidimensional influences of the food environment on dietary behaviors. (Am J Public Health. Published online ahead of print July 16, 2015: e1-e4. doi:10.2105/AJPH.2015.302792).
Article
To examine the relationship between local food environments and obesity and assess the quality of studies reviewed. Systematic keyword searches identified studies from US and Canada that assessed the relationship of obesity to local food environments. We applied a quality metric based on design, exposure and outcome measurement, and analysis. We identified 71 studies representing 65 cohorts. Overall, study quality was low; 60 studies were cross-sectional. Associations between food outlet availability and obesity were predominantly null. Among non-null associations, we saw a trend toward inverse associations between supermarket availability and obesity (22 negative, 4 positive, 67 null) and direct associations between fast food and obesity (29 positive, 6 negative, 71 null) in adults. We saw direct associations between fast food availability and obesity in lower income children (12 positive, 7 null). Indices including multiple food outlets were most consistently associated with obesity in adults (18 expected, 1 not expected, 17 null). Limiting to higher quality studies did not affect results. Despite the large number of studies, we found limited evidence for associations between local food environments and obesity. The predominantly null associations should be interpreted cautiously due to the low quality of available studies. © 2015 The Obesity Society.
Conference Paper
A practical solution to location privacy should be incrementally deployable. We claim it should hence reconcile the economic value of location to aggregators, usually ignored by prior works, with a user's control over her information. Location information indeed is being collected and used by many mobile services to improve revenues, and this gives rise to a heated debate: Privacy advocates ask for stricter regulation on information collection, while companies argue that it would jeopardize the thriving economy of the mobile web. We describe a system that gives users control over their information and does not degrade the data given to aggregators. Recognizing that the first challenge is to express locations in a way that is meaningful for advertisers and users, we propose a keyword-based design. Keywords characterize locations, let the users inform the system about their sensitivity to disclosure, and build information directly usable by an advertiser's targeting campaign. Our work makes two main contributions: we design a market of location information based on keywords and we analyze its robustness to attacks using data from ad-networks, geo-located services, and cell networks.
Article
Geographic treatments of the etiology of obesity tend to turn on the obesogenic environment thesis and investigate the relationship between urban form and obesity. With their emphasis on environmental features that mediate eating and exercise activities, these explorations fundamentally rest on behavioral models of obesogenesis. As such, they tend to black-box the biological body as the site where excess calories are putatively metabolized into fat and made unhealthy. Drawing on critical political ecology, this article discusses the limitations of this dominant approach. First it provides some anomalies not well explained by the energy balance model. Then it reports on emerging biomedical research regarding the role of the endocrine system and endocrine-disrupting chemicals in transforming body ecologies to make them more susceptible to adiposity, regardless of caloric intake. This research also points to the active role of adipose tissue in regulating fat. In light of this evidence, the article argues for a rethinking of current geographical approaches to obesity and health more generally, with due attention to the ecologies of bodies as well as the interpretation of science.
Article
Food access among low-income populations has long been a concern in the United States. Recent research on the geographic dimensions of access has focused on economically deprived areas with little retail food activity, referred to as &apos;food deserts. &apos; We illustrate concepts of urban food access in this descriptive case study from post-Katrina New Orleans. We augment conventional definitions of food deserts by considering a variety of retail food outlets from a complete mapping of the city and by incorporating data on in-store contents on availability and shelf space of fruits and vegetables from a stratified sample of outlets. We show that the existence of food deserts depends on the definitions employed; commonly-used constructs in the food desert literature result in prevalence rates for New Orleans of anywhere from 17 % to 87% of the city&apos;s 175 census tracts. Ambiguities inherent in the construct do not diminish the fact that long travel distances to procure food do increase at-home food costs and that contextual effects on individual health and community development are often associated with areas of impoverished food resources. Describing poor geographic access can improve assessments of household resource inadequacy; we illustrate how transport costs can be used to inform federal food assistance policy. We also show how identifying areas of need can be used at the local level to prioritize retail food projects. Given the current problems of over-nutrition, the paper concludes by suggesting a more useful geographic metaphor would be “food swamps, ” areas in which large relative amounts of energy-dense snack foods, inundate healthy food options. 2
Article
Many people in developed countries fail to consume a healthy diet. This phenomenon has been linked to the contested existence of ‘food deserts’ in the UK, and the occurrence of ‘food insecurity’ in the USA and elsewhere. ‘Food deserts’ remain contested theoretical territory at least partly because no firm definition has been proposed. This paper argues that the barriers to consumption of a healthy diet may be classified according to whether such barriers are financial, physical, or derive from the mental attitude and knowledge of the consumer. The perception of ‘unsupportive food environments’ by some consumers is contrasted with the geographical existence of multiple sources of fresh fruit and vegetables in certain locations. Using a total of 234 semi—structured interviews in various UK locations, qualitative evidence is gathered for the existence of at least ten different types of ‘food desert’. The paper then goes on to show how such a three fold classification may be developed, using a modified ternary diagram, to assess the most appropriate initiatives to tackle ‘food deserts’ and to monitor progress in alleviating their effects.
Article
Abstract  The concept of the food desert, an area with limited access to retail food stores, has increasingly been used within social scientific and public health research to explore the dimensions of spatial inequality and community well-being. While research has demonstrated that food deserts are frequently characterized by higher levels of poverty and food insecurity, there has been relatively little research examining the relationship between food deserts and obesity, particularly in rural areas. In this article we use Geographic Information System (GIS) techniques to identify food desert areas in rural Pennsylvania. We then analyze student body mass index (BMI) data along with census and school district-level data to determine the extent to which the percentage of a school district's population residing within a food desert is positively associated with increased incidence of child overweight among students within the district. We find that school districts with higher percentages of populations located within food deserts are more likely to be structurally and economically disadvantaged. Net of these district-level structural and economic characteristics, we additionally find a positive relationship between increased rates of child overweight and the percentage of the district population residing in a food desert.
Article
Despite growing attention to the problem of obesogenic environments, there has not been a comprehensive review evaluating the food environment-diet relationship. This study aims to evaluate this relationship in the current literature, focusing specifically on the method of exposure assessment (GIS, survey, or store audit). This study also explores 5 dimensions of "food access" (availability, accessibility, affordability, accommodation, acceptability) using a conceptual definition proposed by Penchansky and Thomas (1981). Articles were retrieved through a systematic keyword search in Web of Science and supplemented by the reference lists of included studies. Thirty-eight studies were reviewed and categorized by the exposure assessment method and the conceptual dimensions of access it captured. GIS-based measures were the most common measures, but were less consistently associated with diet than other measures. Few studies examined dimensions of affordability, accommodation, and acceptability. Because GIS-based measures on their own may not capture important non-geographic dimensions of access, a set of recommendations for future researchers is outlined.
Article
Policies on menu labeling have been proposed as a method to improve the food environment. However, there is little information on the nutrient content of chain restaurant menu items and changes over time. To evaluate the energy, saturated fat, and sodium content of entrées 6 and 18 months post-implementation of restaurant menu labeling in King County of Washington State for items that were on the menu at both time periods, and across all items at 6 and 18 months and to compare energy content to recommendations provided by the 2005 Dietary Guidelines for Americans. Eligible restaurants included sit-down and quick-service chains (eg, burgers, pizza, sandwiches/subs, and Tex-Mex) subject to King County regulations with four or more establishments. One establishment per chain was audited at each time period. Hypothesis one examined entrées that were on the menu at both time periods using a paired t test and hypothesis two compared quartiles at 6 months to the distribution at 18 months using a Mantel-Haentzel odds ratios and 95% CIs, and a Cochrane-Armitage test for trend. The content of entrées at 18 months was compared with one-third (assuming three meals per day) of the nutrient intake recommendations for adults provided by the 2005 Dietary Guidelines for Americans. The audit included 37 eligible chains of 92 regulated chains. Energy contents were lower (all chains -41, sit down -73, and quick service -19; paired t tests P<0.0001) for entrées that were on the menu at both time periods. There was a significant trend across quartiles for a decrease in energy, saturated fat, and sodium for all entrées at sit-down chains only. At 18 months entrées not designated for children exceeded 56%, 77%, and 89% of the energy, saturated fat, and sodium guidelines, respectively. Modest improvements in the nutrient content of sit-down and quick-service restaurant entrées occurred but overall levels for energy, saturated fat, and sodium are excessive.
Article
The past decades have seen an increased interest in understanding how the environment affects population health. In particular, public health practitioners and researchers alike are eager to know how the food environments of neighborhoods, schools, and worksites affect food choices and, ultimately, population risk for obesity and other diet-related chronic disease. However, the measurement tools for assessing the environment and the employed study designs have limited our ability to gain important ground. The field has not yet fully considered the psychometric properties of the environmental measurement tools, or how to deal with the copious amounts of data generated from many environmental measures. The field is dominated by research using unsophisticated study designs and has frequently failed to see the role of social and individual factors and how they interrelate with the physical environment. This paper examines some of the measurement issues to be considered as public health practitioners and researchers attempt to understand the impact of the food environment on the health of communities and takes a broad look at where the science currently is with regard to how the food environment is measured, thoughts on what issues may benefit from more deliberate inspection, and directions for future work.
Article
Increasingly, studies are focusing on the role the local food environment plays in residents' ability to purchase affordable, healthy and nutritious foods. In a food desert, an area devoid of a supermarket, access to healthy food is limited. We conducted a systematic review of studies that focused on food access and food desert research in the United States. The 31 studies identified utilized 9 measures to assess food access. Results from these studies can be summarized primarily into four major statements. Findings from other countries offer insight into ways, in which future research, policy development and program implementation in the U.S. may continue to be explored.
Article
The frequent consumption of energy-dense fast food is associated with increased body mass index. This systematic review aims to examine the methodology and current evidence on fast food access and its associations with outcomes. Six databases were searched using terms relating to fast food. Only peer-reviewed studies published in English during a 10-year period, with data collection and analysis regarding fast food access were included. Forty articles met the aforementioned criteria. Nearly half of the studies (n = 16) used their own set of features to define fast food. Studies predominantly examined the relationship between fast food access and socioeconomic factors (n = 21) and 76% indicated fast food restaurants were more prevalent in low-income areas compared with middle- to higher-income areas. Ten of 12 studies found fast food restaurants were more prevalent in areas with higher concentrations of ethnic minority groups in comparison with Caucasians. Six adult studies found higher body mass index was associated with living in areas with increased exposure to fast food; four studies, however, did not find associations. Further work is needed to understand if and how fast food access impacts dietary intake and health outcomes; and if fast food access has disparate socioeconomic, race/ethnicity and age associations.
Article
We conducted a health impact assessment to quantify the potential impact of a state menu-labeling law on population weight gain in Los Angeles County, California. We utilized published and unpublished data to model consumer response to point-of-purchase calorie postings at large chain restaurants in Los Angeles County. We conducted sensitivity analyses to account for uncertainty in consumer response and in the total annual revenue, market share, and average meal price of large chain restaurants in the county. Assuming that 10% of the restaurant patrons would order reduced-calorie meals in response to calorie postings, resulting in an average reduction of 100 calories per meal, we estimated that menu labeling would avert 40.6% of the 6.75 million pound average annual weight gain in the county population aged 5 years and older. Substantially larger impacts would be realized if higher percentages of patrons ordered reduced-calorie meals or if average per-meal calorie reductions increased. Our findings suggest that mandated menu labeling could have a sizable salutary impact on the obesity epidemic, even with only modest changes in consumer behavior.
Article
To examine changes, between 1977-78 and 1994-96, in the quantity and quality of food Americans consumed that was prepared at home versus away from home. Data were obtained from nationwide surveys of food consumption conducted by the US Department of Agriculture (USDA) in 1977-78 and 1994-96. To maximize comparability, we used "day 1" dietary data, which both surveys collected via 24-hour recall. Individuals 2 years of age and over were selected. USDA sampling weights were used to generate nationally representative estimates. We categorized foods by preparation at home or at restaurants, fast-food establishments, schools/day care, and other non-home locations. We assessed percent calories from total fat and saturated fat, and the cholesterol, sodium, fiber, calcium, and iron densities of foods prepared at home versus those prepared away from home. T tests were calculated using accepted procedures to adjust for survey design effects. Between 1977-78 and 1994-96, consumption of food prepared away from home increased from 18% to 32% of total calories. Meals and snacks based on food prepared away from home contained more calories per eating occasion, and "away" food was higher in total fat and saturated fat on a per-calorie basis than at-home food. "Away" food contained less dietary fiber, calcium, and iron on a per-calorie basis. Among adults but not children, food prepared away from home was more sodium and cholesterol dense. When developing intervention messages and strategies, nutrition educators need to be aware of the increasing role of "away" food in Americans' diets.
Article
Is the food business to blame for the fattening of America? Can nutrition professionals play the blame game and get themselves off the hook? Some perspectives on the strategy and tactics of the war against obesity.
Article
The primary objective of this study was to test the comparability of digital photography and visual estimation procedures for estimating food intake. The study sample included 71 breakfast meals and 59 lunch meals eaten in a university cafeteria during a single day. A total of 66 different foods were employed as test foods that could be selected by the students. Food selections and plate waste, as estimated by digital photography and visual estimation, were compared. For digital photography, three observers independently estimated portion sizes of each food item based upon digital photographs. One observer estimated portion sizes in the cafeteria setting, using visual estimation, a method that has been validated in other studies. To test the accuracy of the two procedures for measuring food intake, the estimates of food weights derived from both procedures were compared using Bland-Altman regression. In comparison to visual estimation, the digital photography method yielded comparable estimates of food selections, plate waste, and total food intake for seven of nine comparisons. The two methods of estimating food portions yielded comparable results for most (78%) types of foods. The two methods also yielded similar variability. These findings suggest that the digital photography method is an alternative to the traditional method of estimating food intake via direct observation.
Article
The chapter reviews the historical development and interactions of U.S. agricultural, economic, nutrition, and development policies relating to the creation of the commercial environment of social, economic, technological, and political factors that favor the development of American obesity.
Article
To date, nearly all research on physical activity and the built environment is based on self-reported physical activity and perceived assessment of the built environment. To assess how objectively measured levels of physical activity are related with objectively measured aspects of the physical environment around each participant's home while controlling for sociodemographic covariates. Objective measures of the built environment unique to each household's physical location were developed within a geographic information system to assess land-use mix, residential density, and street connectivity. These measures were then combined into a walkability index. Accelerometers were deployed over a 2-day period to capture objective levels of physical activity in 357 adults. Measures of land-use mix, residential density, and intersection density were positively related with number of minutes of moderate physical activity per day. A combined walkability index of these urban form factors was significant (p =0.002) and explained additional variation in the number of minutes of moderate activity per day over sociodemographic covariates. Thirty-seven percent of individuals in the highest walkability index quartile met the > or =30 minutes of physical activity recommended, compared to only 18% of individuals in the lowest walkability quartile. Individuals in the highest walkability quartile were 2.4 times more likely (confidence interval=1.18-4.88) than individuals in the lowest walkability quartile to meet the recommended > or =30 minutes of moderate physical activity per day. This research supports the hypothesis that community design is significantly associated with moderate levels of physical activity. These results support the rationale for the development of policy that promotes increased levels of land-use mix, street connectivity, and residential density as interventions that can have lasting public health benefits.
Article
Over the past forty years various changes in the U.S. "built environment" have promoted sedentary lifestyles and less healthful diets. James Sallis and Karen Glanz investigate whether these changes have had a direct effect on childhood obesity and whether improvements to encourage more physical activity and more healthful diets are likely to lower rates of childhood obesity. Researchers, say Sallis and Glanz, have found many links between the built environment and children's physical activity, but they have yet to find conclusive evidence that aspects of the built environment promote obesity. For example, certain development patterns, such as a lack of sidewalks, long distances to schools, and the need to cross busy streets, discourage walking and biking to school. Eliminating such barriers can increase rates of active commuting. But researchers cannot yet prove that more active commuting would reduce rates of obesity. Sallis and Glanz note that recent changes in the nutrition environment, including greater reliance on convenience foods and fast foods, a lack of access to fruits and vegetables, and expanding portion sizes, are also widely believed to contribute to the epidemic of childhood obesity. But again, conclusive evidence that changes in the nutrition environment will reduce rates of obesity does not yet exist. Research into the link between the built environment and childhood obesity is still in its infancy. Analysts do not know whether changes in the built environment have increased rates of obesity or whether improvements to the built environment will decrease them. Nevertheless, say Sallis and Glanz, the policy implications are clear. People who have access to safe places to be active, neighborhoods that are walkable, and local markets that offer healthful food are likely to be more active and to eat more healthful food-two types of behavior that can lead to good health and may help avoid obesity.
Article
Americans are increasingly eating out, but nutrition environments in restaurants are poorly understood. An observational measure was developed to assess factors believed to contribute to food choices in restaurants, including availability of more healthy foods, facilitators and barriers to healthful eating, pricing, and signage/promotion of healthy and unhealthy foods. Inter-rater and test-retest reliability were assessed in 217 sit-down and fast-food restaurants in four neighborhoods in 2004 and 2005. Inter-rater reliability was generally high, with most kappa values greater than 0.80 (range 0.27-0.97) and all percent-agreement values greater than 75% (77.6-99.5). Test-retest reliability was high, with most kappa values greater than 0.80 (0.46-1.0) and all percent-agreement values greater than 80% (80.4-100). There were several differences (p<0.05) between nutrition environment variables in sit-down versus fast-food restaurants, although neither restaurant type was consistently more healthful. Fast-food restaurants had greater healthy entrée and main-dish salad availability, but sit-down restaurants had a higher proportion of healthy main-dish salads and more healthy food and beverage items. Fast-food restaurants more often encouraged large portions, unhealthful eating, and overeating, and offered relative cost savings for combination meals, but were also more likely to provide nutrition information and highlight healthy options. Testing hypotheses about food environment influences on obesity and eating patterns requires psychometrically sound measurement of nutrition environments. This Nutrition Environment Measures Study restaurant assessment (NEMS-R) has evidence of reliability, and can discriminate restaurant types. The NEMS-R can be used in research and practice to characterize restaurant environments.
Article
Research has shown that lower socioeconomic groups purchase foods that are less consistent with dietary recommendations. The price and availability of foods are thought to be important mediating factors between socioeconomic position and food purchasing. We examined the relative contribution of the perceived and objectively measured price and availability of recommended foods to household income differences in food purchasing. Using a face-to-face interview, a random sample of Brisbane residents (n=812) were asked about their food purchasing choices in 2000. They were also asked about their perceptions of the price and availability of 'recommended' foods (i.e. choices lower in fat, saturated fat, sugar, salt or higher in fibre) in the supermarkets where they usually shopped. Audits measuring the actual availability and price of identical foods were conducted in the same supermarkets. Lower socioeconomic groups were less likely to make food purchasing choices consistent with dietary guideline recommendations. Objective availability and price differences were not associated with purchasing choices, nor did they contribute to socioeconomic inequalities in food purchasing choices. Perceived availability and price differences were associated with the purchase of recommended foods. Perceived availability made a small contribution to inequalities in food purchasing, however perceived price differences did not. Socioeconomic inequalities in food purchasing are not mediated by differential availability of recommended foods and differences in price between recommended and regular foods in supermarkets, or by perceptions of their relative price. However, differential perceptions of the availability of recommended foods may play a small role in food purchasing inequalities.
Article
This review critically summarizes the literature on neighborhood determinants of obesity and proposes a conceptual framework to guide future inquiry. Thirty-seven studies met all inclusion criteria and revealed that the influence of neighborhood-level factors appears mixed. Neighborhood-level measures of economic resources were associated with obesity in 15 studies, while the associations between neighborhood income inequality and racial composition with obesity were mixed. Availability of healthy versus unhealthy food was inconsistently related to obesity, while neighborhood features that discourage physical activity were consistently associated with increased body mass index. Theoretical explanations for neighborhood-obesity effects and recommendations for strengthening the literature are presented.