How has big data contributed to obesity research? A review of the literature

Advancing precision public health for obesity in children

Worldwide, far too many children and adolescents are living with the disease of obesity. Despite decades of public health initiatives, rates are still rising in many countries. This raises the question of whether precision public health may be a more successful approach to preventing obesity in youth. In this review, the objective was to review the literature on precision public health in the context of childhood obesity prevention and to discuss how precision public health may advance the field of childhood obesity prevention. As precision public health is a concept that is still evolving and not fully identifiable in the literature, a lack of published studies precluded a formal review. Therefore, the approach of using a broad interpretation of precision public health was used and recent advances in childhood obesity research in the areas of surveillance and risk factor identification as well as intervention, evaluation and implementation using selected studies were summarized. Encouragingly, big data from a multitude of designed and organic sources are being used in new and innovative ways to provide more granular surveillance and risk factor identification in obesity in children. Challenges were identified in terms of data access, completeness, and integration, ensuring inclusion of all members of society, ethics, and translation to policy. As precision public health advances, it may yield novel insights that can contribute to strong policies acting in concert that ultimately lead to the prevention of obesity in children.

Non-traditional data sources in obesity research: a systematic review of their use in the study of obesogenic environments

BACKGROUND

The complex nature of obesity increasingly requires a comprehensive approach that includes the role of environmental factors. For understanding contextual determinants, the resources provided by technological advances could become a key factor in obesogenic environment research. This study aims to identify different sources of non-traditional data and their applications, considering the domains of obesogenic environments: physical, sociocultural, political and economic.

METHODS

We conducted a systematic search in PubMed, Scopus and LILACS databases by two independent groups of reviewers, from September to December 2021. We included those studies oriented to adult obesity research using non-traditional data sources, published in the last 5 years in English, Spanish or Portuguese. The overall reporting followed the PRISMA guidelines.

RESULTS

The initial search yielded 1583 articles, 94 articles were kept for full-text screening, and 53 studies met the eligibility criteria and were included. We extracted information about countries of origin, study design, observation units, obesity-related outcomes, environment variables, and non-traditional data sources used. Our results revealed that most of the studies originated from high-income countries (86.54%) and used geospatial data within a GIS (76.67%), social networks (16.67%), and digital devices (11.66%) as data sources. Geospatial data were the most utilised data source and mainly contributed to the study of the physical domains of obesogenic environments, followed by social networks providing data to the analysis of the sociocultural domain. A gap in the literature exploring the political domain of environments was also evident.

CONCLUSION

The disparities between countries are noticeable. Geospatial and social network data sources contributed to studying the physical and sociocultural environments, which could be a valuable complement to those traditionally used in obesity research. We propose the use of information available on the Internet, addressed by artificial intelligence-based tools, to increase the knowledge on political and economic dimensions of the obesogenic environment.

Applications of Artificial Intelligence to Obesity Research: Scoping Review of Methodologies

BACKGROUND

Obesity is a leading cause of preventable death worldwide. Artificial intelligence (AI), characterized by machine learning (ML) and deep learning (DL), has become an indispensable tool in obesity research.

OBJECTIVE

This scoping review aimed to provide researchers and practitioners with an overview of the AI applications to obesity research, familiarize them with popular ML and DL models, and facilitate the adoption of AI applications.

METHODS

We conducted a scoping review in PubMed and Web of Science on the applications of AI to measure, predict, and treat obesity. We summarized and categorized the AI methodologies used in the hope of identifying synergies, patterns, and trends to inform future investigations. We also provided a high-level, beginner-friendly introduction to the core methodologies to facilitate the dissemination and adoption of various AI techniques.

RESULTS

We identified 46 studies that used diverse ML and DL models to assess obesity-related outcomes. The studies found AI models helpful in detecting clinically meaningful patterns of obesity or relationships between specific covariates and weight outcomes. The majority (18/22, 82%) of the studies comparing AI models with conventional statistical approaches found that the AI models achieved higher prediction accuracy on test data. Some (5/46, 11%) of the studies comparing the performances of different AI models revealed mixed results, indicating the high contingency of model performance on the data set and task it was applied to. An accelerating trend of adopting state-of-the-art DL models over standard ML models was observed to address challenging computer vision and natural language processing tasks. We concisely introduced the popular ML and DL models and summarized their specific applications in the studies included in the review.

CONCLUSIONS

This study reviewed AI-related methodologies adopted in the obesity literature, particularly ML and DL models applied to tabular, image, and text data. The review also discussed emerging trends such as multimodal or multitask AI models, synthetic data generation, and human-in-the-loop that may witness increasing applications in obesity research.

Machine learning to predict in-hospital mortality among patients with severe obesity: Proof of concept study

Objectives

Hospitalized patients with severe obesity require adapted hospital management. The aim of this study was to evaluate a machine learning model to predict in-hospital mortality among this population.

Methods

Data of unselected consecutive emergency department admissions of hospitalized patients with severe obesity (BMI ≥ 40 kg/m2) was analyzed. Data was retrieved from five hospitals from the Mount Sinai health system, New York. The study time frame was between January 2011 and December 2019. Data was used to train a gradient-boosting machine learning model to identify in-hospital mortality. The model was trained and evaluated based on the data from four hospitals and externally validated on held-out data from the fifth hospital.

Results

A total of 14,078 hospital admissions of inpatients with severe obesity were included. The in-hospital mortality rate was 297/14,078 (2.1%). In univariate analysis, albumin (area under the curve [AUC] = 0.77), blood urea nitrogen (AUC = 0.76), acuity level (AUC = 0.73), lactate (AUC = 0.72), and chief complaint (AUC = 0.72) were the best single predictors. For Youden's index, the model had a sensitivity of 0.77 (95% CI: 0.67-0.86) with a false positive rate of 1:9.

Conclusion

A machine learning model trained on clinical measures provides proof of concept performance in predicting mortality in patients with severe obesity. This implies that such models may help to adopt specific decision support tools for this population.

Food Sales and Adult Weight Status: Results of a Cross-Sectional Study in England

Ecological studies often use supermarket location as a proxy measure of the food environment. In this study, we used data on sales at a leading mainstream supermarket chain to explore how area-level supermarket use is associated with overweight and obesity in English adults. Sales data were aggregated to local authority level and joined to a national dataset describing self-reported height and weight and fruit and vegetable consumption. Regression models showed a modest association between higher levels of unhealthy food sales relative to health food sales and increased odds of being overweight and obese. Although effect sizes were small, they persisted after adjustment for area-level deprivation. Supermarket sales data provide additional understanding in the study of food environments and their impact on increasing weight status. Future health policies should consider using ‘big data’ combined with other research methods to address the increasing consumption of unhealthy and highly processed foods.

Traditional Machine Learning Models and Bidirectional Encoder Representations From Transformer (BERT)-Based Automatic Classification of Tweets About Eating Disorders: Algorithm Development and Validation Study

BACKGROUND

Eating disorders affect an increasing number of people. Social networks provide information that can help.

OBJECTIVE

We aimed to find machine learning models capable of efficiently categorizing tweets about eating disorders domain.

METHODS

We collected tweets related to eating disorders, for 3 consecutive months. After preprocessing, a subset of 2000 tweets was labeled: (1) messages written by people suffering from eating disorders or not, (2) messages promoting suffering from eating disorders or not, (3) informative messages or not, and (4) scientific or nonscientific messages. Traditional machine learning and deep learning models were used to classify tweets. We evaluated accuracy, F1 score, and computational time for each model.

RESULTS

A total of 1,058,957 tweets related to eating disorders were collected. were obtained in the 4 categorizations, with The bidirectional encoder representations from transformer-based models had the best score among the machine learning and deep learning techniques applied to the 4 categorization tasks (F1 scores 71.1%-86.4%).

CONCLUSIONS

Bidirectional encoder representations from transformer-based models have better performance, although their computational cost is significantly higher than those of traditional techniques, in classifying eating disorder-related tweets.

A systematic review of supermarket automated electronic sales data for population dietary surveillance

Context

Most dietary assessment methods are limited by self-report biases, how long they take for participants to complete, and cost of time for dietitians to extract content. Electronically recorded, supermarket-obtained transactions are an objective measure of food purchases, with reduced bias and improved timeliness and scale.

Objective

The use, breadth, context, and utility of electronic purchase records for dietary research is assessed and discussed in this systematic review.

Data sources

Four electronic databases (MEDLINE, EMBASE, PsycINFO, Global Health) were searched. Included studies used electronically recorded supermarket transactions to investigate the diet of healthy, free-living adults.

Data extraction

Searches identified 3422 articles, of which 145 full texts were retrieved and 72 met inclusion criteria. Study quality was assessed using the National Institutes of Health Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies.

Data analysis

Purchase records were used in observational studies, policy evaluations, and experimental designs. Nutrition outcomes included dietary patterns, nutrients, and food category sales. Transactions were linked to nutrient data from retailers, commercial data sources, and national food composition databases.

Conclusion

Electronic sales data have the potential to transform dietary assessment and worldwide understanding of dietary behavior. Validation studies are warranted to understand limits to agreement and extrapolation to individual-level diets.

Systematic Review Registration

PROSPERO registration no. CRD42018103470

An ensemble-based feature selection framework to select risk factors of childhood obesity for policy decision making

BACKGROUND

The increasing prevalence of childhood obesity makes it essential to study the risk factors with a sample representative of the population covering more health topics for better preventive policies and interventions. It is aimed to develop an ensemble feature selection framework for large-scale data to identify risk factors of childhood obesity with good interpretability and clinical relevance.

METHODS

We analyzed the data collected from 426,813 children under 18 during 2000-2019. A BMI above the 90th percentile for the children of the same age and gender was defined as overweight. An ensemble feature selection framework, Bagging-based Feature Selection framework integrating MapReduce (BFSMR), was proposed to identify risk factors. The framework comprises 5 models (filter with mutual information/SVM-RFE/Lasso/Ridge/Random Forest) from filter, wrapper, and embedded feature selection methods. Each feature selection model identified 10 variables based on variable importance. Considering accuracy, F-score, and model characteristics, the models were classified into 3 levels with different weights: Lasso/Ridge, Filter/SVM-RFE, and Random Forest. The voting strategy was applied to aggregate the selected features, with both feature weights and model weights taken into consideration. We compared our voting strategy with another two for selecting top-ranked features in terms of 6 dimensions of interpretability.

RESULTS

Our method performed the best to select the features with good interpretability and clinical relevance. The top 10 features selected by BFSMR are age, sex, birth year, breastfeeding type, smoking habit and diet-related knowledge of both children and mothers, exercise, and Mother's systolic blood pressure.

CONCLUSION

Our framework provides a solution for identifying a diverse and interpretable feature set without model bias from large-scale data, which can help identify risk factors of childhood obesity and potentially some other diseases for future interventions or policies.

Understanding Barriers to Novel Data Linkages: Topic Modeling of the Results of the LifeInfo Survey

Background

Novel consumer and lifestyle data, such as those collected by supermarket loyalty cards or mobile phone exercise tracking apps, offer numerous benefits for researchers seeking to understand diet- and exercise-related risk factors for diseases. However, limited research has addressed public attitudes toward linking these data with individual health records for research purposes. Data linkage, combining data from multiple sources, provides the opportunity to enhance preexisting data sets to gain new insights.

Objective

The aim of this study is to identify key barriers to data linkage and recommend safeguards and procedures that would encourage individuals to share such data for potential future research.

Methods

The LifeInfo Survey consulted the public on their attitudes toward sharing consumer and lifestyle data for research purposes. Where barriers to data sharing existed, participants provided unstructured survey responses detailing what would make them more likely to share data for linkage with their health records in the future. The topic modeling technique latent Dirichlet allocation was used to analyze these textual responses to uncover common thematic topics within the texts.

Results

Participants provided responses related to sharing their store loyalty card data (n=2338) and health and fitness app data (n=1531). Key barriers to data sharing identified through topic modeling included data safety and security, personal privacy, requirements of further information, fear of data being accessed by others, problems with data accuracy, not understanding the reason for data linkage, and not using services that produce these data. We provide recommendations for addressing these issues to establish the best practice for future researchers interested in using these data.

Conclusions

This study formulates a large-scale consultation of public attitudes toward this kind of data linkage, which is an important first step in understanding and addressing barriers to participation in research using novel consumer and lifestyle data.

Progress Towards Using Linked Population-Based Data For Geohealth Research: Comparisons Of Aotearoa New Zealand And The United Kingdom

Globally, geospatial concepts are becoming increasingly important in epidemiological and public health research. Individual level linked population-based data afford researchers with opportunities to undertake complex analyses unrivalled by other sources. However, there are significant challenges associated with using such data for impactful geohealth research. Issues range from extracting, linking and anonymising data, to the translation of findings into policy whilst working to often conflicting agendas of government and academia. Innovative organisational partnerships are therefore central to effective data use. To extend and develop existing collaborations between the institutions, in June 2019, authors from the Leeds Institute for Data Analytics and the Alan Turing Institute, London, visited the Geohealth Laboratory based at the University of Canterbury, New Zealand. This paper provides an overview of insight shared during a two-day workshop considering aspects of linked population-based data for impactful geohealth research. Specifically, we discuss both the collaborative partnership between New Zealand's Ministry of Health (MoH) and the University of Canterbury's GeoHealth Lab and novel infrastructure, and commercial partnerships enabled through the Leeds Institute for Data Analytics and the Alan Turing Institute in the UK. We consider the New Zealand Integrated Data Infrastructure as a case study approach to population-based linked health data and compare similar approaches taken by the UK towards integrated data infrastructures, including the ESRC Big Data Network centres, the UK Biobank, and longitudinal cohorts. We reflect on and compare the geohealth landscapes in New Zealand and the UK to set out recommendations and considerations for this rapidly evolving discipline.

Smart city lifestyle sensing, big data, geo-analytics and intelligence for smarter public health decision-making in overweight, obesity and type 2 diabetes prevention: the research we should be doing

The public health burden caused by overweight, obesity (OO) and type-2 diabetes (T2D) is very significant and continues to rise worldwide. The causation of OO and T2D is complex and highly multifactorial rather than a mere energy intake (food) and expenditure (exercise) imbalance. But previous research into food and physical activity (PA) neighbourhood environments has mainly focused on associating body mass index (BMI) with proximity to stores selling fresh fruits and vegetables or fast food restaurants and takeaways, or with neighbourhood walkability factors and access to green spaces or public gym facilities, making largely naive, crude and inconsistent assumptions and conclusions that are far from the spirit of 'precision and accuracy public health'. Different people and population groups respond differently to the same food and PA environments, due to a myriad of unique individual and population group factors (genetic/epigenetic, metabolic, dietary and lifestyle habits, health literacy profiles, screen viewing times, stress levels, sleep patterns, environmental air and noise pollution levels, etc.) and their complex interplays with each other and with local food and PA settings. Furthermore, the same food store or fast food outlet can often sell or serve both healthy and non-healthy options/portions, so a simple binary classification into 'good' or 'bad' store/outlet should be avoided. Moreover, appropriate physical exercise, whilst essential for good health and disease prevention, is not very effective for weight maintenance or loss (especially when solely relied upon), and cannot offset the effects of a bad diet. The research we should be doing in the third decade of the twenty-first century should use a systems thinking approach, helped by recent advances in sensors, big data and related technologies, to investigate and consider all these factors in our quest to design better targeted and more effective public health interventions for OO and T2D control and prevention.

A Systematic Review of Economic Evaluations of Health-Promoting Food Retail-Based Interventions

BACKGROUND

While the number of retail interventions with impacts on diet- and/or health-related outcomes is increasing, the economic evaluation literature is limited. This review investigated (i) the cost-effectiveness of health-promoting food retail interventions and (ii) key assumptions adopted in these evaluations.

METHODS

A systematic review of published academic studies was undertaken (CRD42020153763). Fourteen databases were searched. Eligible studies were identified, analysed, and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines.

RESULTS

Eight studies that evaluated 30 retail interventions were included in the review. Common outcomes reported were cost per healthy food item purchased/served or cost per disability-adjusted life year (DALY) averted. Four studies undertook cost-utility analyses and half of these studies concluded that retail interventions were cost-effective in improving health outcomes. Most studies did not state any assumptions regarding compensatory behaviour (i.e., purchases/consumption of non-intervention foods or food purchases/consumption from non-intervention settings) and presumed that sales data were indicative of consumption.

CONCLUSION

The cost-effectiveness of retail-based health-promoting interventions is inconclusive. Future health-promoting retail interventions should regularly include an economic evaluation which addresses key assumptions related to compensatory behaviour and the use of sales data as a proxy for consumption.

Emerging Applications of Machine Learning in Food Safety

Food safety continues to threaten public health. Machine learning holds potential in leveraging large, emerging data sets to improve the safety of the food supply and mitigate the impact of food safety incidents. Foodborne pathogen genomes and novel data streams, including text, transactional, and trade data, have seen emerging applications enabled by a machine learning approach, such as prediction of antibiotic resistance, source attribution of pathogens, and foodborne outbreak detection and risk assessment. In this article, we provide a gentle introduction to machine learning in the context of food safety and an overview of recent developments and applications. With many of these applications still in their nascence, general and domain-specific pitfalls and challenges associated with machine learning have begun to be recognized and addressed, which are critical to prospective use and future deployment of large data sets and their associated machine learning models for food safety applications.

Open data products-A framework for creating valuable analysis ready data

This paper develops the notion of "open data product". We define an open data product as the open result of the processes through which a variety of data (open and not) are turned into accessible information through a service, infrastructure, analytics or a combination of all of them, where each step of development is designed to promote open principles. Open data products are born out of a (data) need and add value beyond simply publishing existing datasets. We argue that the process of adding value should adhere to the principles of open (geographic) data science, ensuring openness, transparency and reproducibility. We also contend that outreach, in the form of active communication and dissemination through dashboards, software and publication are key to engage end-users and ensure societal impact. Open data products have major benefits. First, they enable insights from highly sensitive, controlled and/or secure data which may not be accessible otherwise. Second, they can expand the use of commercial and administrative data for the public good leveraging on their high temporal frequency and geographic granularity. We also contend that there is a compelling need for open data products as we experience the current data revolution. New, emerging data sources are unprecedented in temporal frequency and geographical resolution, but they are large, unstructured, fragmented and often hard to access due to privacy and confidentiality concerns. By transforming raw (open or "closed") data into ready to use open data products, new dimensions of human geographical processes can be captured and analysed, as we illustrate with existing examples. We conclude by arguing that several parallels exist between the role that open source software played in enabling research on spatial analysis in the 90 s and early 2000s, and the opportunities that open data products offer to unlock the potential of new forms of (geo-)data.

Establishing consensus on key public health indicators for the monitoring and evaluating childhood obesity interventions: a Delphi panel study

BACKGROUND

Childhood obesity is influenced by myriad individual, societal and environmental factors that are not typically reflected in current interventions. Socio-ecological conditions evolve and require ongoing monitoring in terms of assessing their influence on child health. The aim of this study was to identify and prioritise indicators deemed relevant by public health authorities for monitoring and evaluating childhood obesity interventions.

METHOD

A three-round Delphi Panel composed of experts from regions across Europe, with a remit in childhood obesity intervention, were asked to identify indicators that were a priority in their efforts to address childhood obesity in their respective jurisdictions. In Round 1, 16 panellists answered a series of open-ended questions to identify the most relevant indicators concerning the evaluation and subsequent monitoring of interventions addressing childhood obesity, focusing on three main domains: built environments, dietary environments, and health inequalities. In Rounds 2 and 3, panellists rated the importance of each of the identified indicators within these domains, and the responses were then analysed quantitatively.

RESULTS

Twenty-seven expert panellists were invited to participate in the study. Of these, 16/27 completed round 1 (5 9% response rate), 14/16 completed round 2 (87.5% response rate), and 8/14 completed the third and final round (57% response rate). Consensus (defined as > 70% agreement) was reached on a total of 45 of the 87 indicators (49%) across three primary domains (built and dietary environments and health inequalities), with 100% consensus reached for 5 of these indicators (6%).

CONCLUSION

Forty-five potential indicators were identified, pertaining primarily to the dietary environment, built environment and health inequalities. These results have important implications more widely for evaluating interventions aimed at childhood obesity reduction and prevention.

Big Data Applications in Medical Field: A Literature Review

Digital imaging and medical reporting have acquired an essential role in healthcare, but the main challenge is the storage of a high volume of patient data. Although newer technologies are already introduced in the medical sciences to save records size, Big Data provides advancements by storing a large amount of data to improve the efficiency and quality of patient treatment with better care. It provides intelligent automation capabilities to reduce errors than manual inputs. Large numbers of research papers on big data in the medical field are studied and analyzed for their impacts, benefits, and applications. Big data has great potential to support the digitalization of all medical and clinical records and then save the entire data regarding the medical history of an individual or a group. This paper discusses big data usage for various industries and sectors. Finally, 12 significant applications for the medical field by the implementation of big data are identified and studied with a brief description. This technology can be gainfully used to extract useful information from the available data by analyzing and managing them through a combination of hardware and software. With technological advancement, big data provides health-related information for millions of patient-related to life issues such as lab tests reporting, clinical narratives, demographics, prescription, medical diagnosis, and related documentation. Thus, Big Data is essential in developing a better yet efficient analysis and storage healthcare services. The demand for big data applications is increasing due to its capability of handling and analyzing massive data. Not only in the future but even now, Big Data is proving itself as an axiom of storing, developing, analyzing, and providing overall health information to the physicians.

The Academic Viewpoint on Patient Data Ownership in the Context of Big Data: Scoping Review

BACKGROUND

The ownership of patient information in the context of big data is a relatively new problem, which is not yet fully recognized by the medical academic community. The problem is interdisciplinary, incorporating legal, ethical, medical, and aspects of information and communication technologies, requiring a sophisticated analysis. However, no previous scoping review has mapped existing studies on the subject.

OBJECTIVE

This study aims to map and assess published studies on patient data ownership in the context of big data as viewed by the academic community.

METHODS

A scoping review was conducted based on the 5-stage framework outlined by Arksey and O'Malley and further developed by Levac, Colquhoun, and O'Brien. The organization and reporting of results of the scoping review were conducted according to PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses and its extensions for Scoping Reviews). A systematic and comprehensive search of 4 scientific information databases, PubMed, ScienceDirect, Scopus, and Springer, was performed for studies published between January 2000 and October 2019. Two authors independently assessed the eligibility of the studies and the extracted data.

RESULTS

The review included 32 eligible articles authored by academicians that correspond to 3 focus areas: problem (ownership), area (health care), and context (big data). Five major aspects were studied: the scientific area of publications, aspects and academicians' perception of ownership in the context of big data, proposed solutions, and practical applications for data ownership issues in the context of big data. The aspects in which publications consider ownership of medical data are not clearly distinguished but can be summarized as ethical, legal, political, and managerial. The ownership of patient data is perceived primarily as a challenge fundamental to conducting medical research, including data sales and sharing, and to a lesser degree as a means of control, problem, threat, and opportunity also in view of medical research. Although numerous solutions falling into 3 categories, technology, law, and policy, were proposed, only 3 real applications were discussed.

CONCLUSIONS

The issue of ownership of patient information in the context of big data is poorly researched; it is not addressed consistently and in its integrity, and there is no consensus on policy decisions and the necessary legal regulations. Future research should investigate the issue of ownership as a core research question and not as a minor fragment among other topics. More research is needed to increase the body of knowledge regarding the development of adequate policies and relevant legal frameworks in compliance with ethical standards. The combined efforts of multidisciplinary academic teams are needed to overcome existing gaps in the perception of ownership, the aspects of ownership, and the possible solutions to patient data ownership issues in the reality of big data.

Comparing supermarket loyalty card data with traditional diet survey data for understanding how protein is purchased and consumed in older adults for the UK, 2014-16

Background

Our ability to understand population-level dietary intake patterns is dependent on having access to high quality data. Diet surveys are common diet assessment methods, but can be limited by bias associated with under-reporting. Food purchases tracked using supermarket loyalty card records may supplement traditional surveys, however they are rarely available to academics and policy makers. The aim of our study is to explore population level patterns of protein purchasing and consumption in ageing adults (40 years onwards).

Methods

We used diet survey data from the National Diet and Nutrition Survey (2014–16) on food consumption, and loyalty card records on food purchases from a major high street supermarket retailer (2016–17) covering the UK. We computed the percentage of total energy derived from protein, protein intake per kg of body mass, and percentage of protein acquired by food type.

Results

We found that protein consumption (as the percentage of total energy purchased) increased between ages 40–65 years, and declined thereafter. In comparison, protein purchased in supermarkets was roughly 2–2.5 percentage points lower at each year of age. The proportion of adults meeting recommended levels of protein was lowest in age groups 55–69 and 70+. The time of protein consumption was skewed towards evening meals, with low intakes during breakfast or between main meals. Meat, fish and poultry dominated as sources of protein purchased and consumed, although adults also acquired a large share of their protein from dairy and bread, with little from plant protein.

Conclusions

Our study provides novel insights into how protein is purchased and consumed by ageing adults in the UK. Supermarket loyalty card data can reveal patterns of protein purchasing that when combined with traditional sources of dietary intake may enhance our understanding of dietary behaviours.

Identification of temporal condition patterns associated with pediatric obesity incidence using sequence mining and big data

BACKGROUND

Electronic health records (EHRs) are potentially important components in addressing pediatric obesity in clinical settings and at the population level. This work aims to identify temporal condition patterns surrounding obesity incidence in a large pediatric population that may inform clinical care and childhood obesity policy and prevention efforts.

METHODS

EHR data from healthcare visits with an initial record of obesity incidence (index visit) from 2009 through 2016 at the Children's Hospital of Philadelphia, and visits immediately before (pre-index) and after (post-index), were compared with a matched control population of patients with a healthy weight to characterize the prevalence of common diagnoses and condition trajectories. The study population consisted of 49,694 patients with pediatric obesity and their corresponding matched controls. The SPADE algorithm was used to identify common temporal condition patterns in the case population. McNemar's test was used to assess the statistical significance of pattern prevalence differences between the case and control populations.

RESULTS

SPADE identified 163 condition patterns that were present in at least 1% of cases; 80 were significantly more common among cases and 45 were significantly more common among controls (p < 0.05). Asthma and allergic rhinitis were strongly associated with childhood obesity incidence, particularly during the pre-index and index visits. Seven conditions were commonly diagnosed for cases exclusively during pre-index visits, including ear, nose, and throat disorders and gastroenteritis.

CONCLUSIONS

The novel application of SPADE on a large retrospective dataset revealed temporally dependent condition associations with obesity incidence. Allergic rhinitis and asthma had a particularly high prevalence during pre-index visits. These conditions, along with those exclusively observed during pre-index visits, may represent signals of future obesity. While causation cannot be inferred from these associations, the temporal condition patterns identified here represent hypotheses that can be investigated to determine causal relationships in future obesity research.

A critical review of diet-related surveys in England, 1970-2018

Background

Many diet-related surveys have been conducted in England over the past four to five decades. Yet, diet-related ill-health is estimated to cost the NHS £5.8 billion annually. There has been no recent assessment of the diet-related surveys currently available in England. This paper aims to fill this gap in the literature by providing researchers, especially those interested in conducting secondary (quantitative) research on diet, with a detailed overview of the major repeated cross-sectional and longitudinal surveys conducted in England over the last 48 years (1970–2018).

Method

A three-stage review process was used to identify and assess surveys and synthesise the information necessary for achieving the paper’s aim. Surveys were identified using the UK Data Service, Cohort and Longitudinal Studies Enhancement Resources (CLOSER), the Medical Research Council (MRC) Cohort Directory and the Consumer Data Research Centre (CDRC) online data repositories/directories. Surveys were summarised to include a brief background, the survey design and methodology used, variables captured, the target population, level of geography covered, the type of dietary assessment method(s) used, primary data users, data accessibility, availability and costs, as well as key survey features and considerations.

Results

The key considerations identified across the various surveys following the review include: the overall survey design and the different dietary assessment method(s) used in each survey; methodological changes and general inconsistencies in the type and quantity of diet-related questions posed across and within surveys over time; and differences in the level of geography and target groups captured.

Conclusion

It is highly unlikely that any survey dataset will meet all the needs of researchers. Nevertheless, researchers are encouraged to make good use of the secondary data currently available, in order to conduct the research necessary for the creation of more evidence-based diet-related policies and interventions in England. The review process used in this paper is one that can be easily replicated and one which future studies can use to update and expand upon to assist researchers in identifying the survey(s) most aligned to their research questions.

Sex and gender differences and biases in artificial intelligence for biomedicine and healthcare

Precision Medicine implies a deep understanding of inter-individual differences in health and disease that are due to genetic and environmental factors. To acquire such understanding there is a need for the implementation of different types of technologies based on artificial intelligence (AI) that enable the identification of biomedically relevant patterns, facilitating progress towards individually tailored preventative and therapeutic interventions. Despite the significant scientific advances achieved so far, most of the currently used biomedical AI technologies do not account for bias detection. Furthermore, the design of the majority of algorithms ignore the sex and gender dimension and its contribution to health and disease differences among individuals. Failure in accounting for these differences will generate sub-optimal results and produce mistakes as well as discriminatory outcomes. In this review we examine the current sex and gender gaps in a subset of biomedical technologies used in relation to Precision Medicine. In addition, we provide recommendations to optimize their utilization to improve the global health and disease landscape and decrease inequalities.

Evidence from big data in obesity research: international case studies

BACKGROUND/OBJECTIVE

Obesity is thought to be the product of over 100 different factors, interacting as a complex system over multiple levels. Understanding the drivers of obesity requires considerable data, which are challenging, costly and time-consuming to collect through traditional means. Use of 'big data' presents a potential solution to this challenge. Big data is defined by Delphi consensus as: always digital, has a large sample size, and a large volume or variety or velocity of variables that require additional computing power (Vogel et al. Int J Obes. 2019). 'Additional computing power' introduces the concept of big data analytics. The aim of this paper is to showcase international research case studies presented during a seminar series held by the Economic and Social Research Council (ESRC) Strategic Network for Obesity in the UK. These are intended to provide an in-depth view of how big data can be used in obesity research, and the specific benefits, limitations and challenges encountered.

METHODS AND RESULTS

Three case studies are presented. The first investigated the influence of the built environment on physical activity. It used spatial data on green spaces and exercise facilities alongside individual-level data on physical activity and swipe card entry to leisure centres, collected as part of a local authority exercise class initiative. The second used a variety of linked electronic health datasets to investigate associations between obesity surgery and the risk of developing cancer. The third used data on tax parcel values alongside data from the Seattle Obesity Study to investigate sociodemographic determinants of obesity in Seattle.

CONCLUSIONS

The case studies demonstrated how big data could be used to augment traditional data to capture a broader range of variables in the obesity system. They also showed that big data can present improvements over traditional data in relation to size, coverage, temporality, and objectivity of measures. However, the case studies also encountered challenges or limitations; particularly in relation to hidden/unforeseen biases and lack of contextual information. Overall, despite challenges, big data presents a relatively untapped resource that shows promise in helping to understand drivers of obesity.

Creating a long-term future for big data in obesity research

Big data are part of the future in obesity research. The ESRC funded Strategic Network for Obesity has together generated a series of papers, published in the International Journal for Obesity illustrating various aspects of their utility, in particular relating to the large social and environmental drivers of obesity. This article is the final part of the series and reflects upon progress to date and identifies four areas that require attention to promote the continued role of big data in research. We additionally include a 'getting started with big data' checklist to encourage more obesity researchers to engage with alternative data resources.

An Area-Level Indicator of Latent Soda Demand: Spatial Statistical Modeling of Grocery Store Transaction Data to Characterize the Nutritional Landscape in Montreal, Canada

Measurement of neighborhood dietary patterns at high spatial resolution allows public health agencies to identify and monitor communities with an elevated risk of nutrition-related chronic diseases. Currently, data on diet are obtained primarily through nutrition surveys, which produce measurements at low spatial resolutions. The availability of store-level grocery transaction data provides an opportunity to refine the measurement of neighborhood dietary patterns. We used these data to develop an indicator of area-level latent demand for soda in the Census Metropolitan Area of Montreal in 2012 by applying a hierarchical Bayesian spatial model to data on soda sales from 1,097 chain retail food outlets. The utility of the indicator of latent soda demand was evaluated by assessing its association with the neighborhood relative risk of prevalent type 2 diabetes mellitus. The indicator improved the fit of the disease-mapping model (deviance information criterion: 2,140 with the indicator and 2,148 without) and enables a novel approach to nutrition surveillance.

A Delphi study to build consensus on the definition and use of big data in obesity research

BACKGROUND

'Big data' has great potential to help address the global health challenge of obesity. However, lack of clarity with regard to the definition of big data and frameworks for effectively using big data in the context of obesity research may be hindering progress. The aim of this study was to establish agreed approaches for the use of big data in obesity-related research.

METHODS

A Delphi method of consensus development was used, comprising three survey rounds. In Round 1, participants were asked to rate agreement/disagreement with 77 statements across seven domains relating to definitions of, and approaches to, using big data in the context of obesity research. Participants were also asked to contribute further ideas in relation to these topics, which were incorporated as new statements (n = 8) in Round 2. In Rounds 2 and 3 participants re-appraised their ratings in view of the group consensus.

RESULTS

Ninety-six experts active in obesity-related research were invited to participate. Of these, 36/96 completed Round 1 (37.5% response rate), 29/36 completed Round 2 (80.6% response rate) and 26/29 completed Round 3 (89.7% response rate). Consensus (defined as > 70% agreement) was achieved for 90.6% (n = 77) of statements, with 100% consensus achieved for the Definition of Big Data, Data Governance, and Quality and Inference domains.

CONCLUSIONS

Experts agreed that big data was more nuanced than the oft-cited definition of 'volume, variety and velocity', and includes quantitative, qualitative, observational or intervention data from a range of sources that have been collected for research or other purposes. Experts repeatedly called for third party action, for example to develop frameworks for reporting and ethics, to clarify data governance requirements, to support training and skill development and to facilitate sharing of big data. Further advocacy will be required to encourage organisations to adopt these roles.

K-means clustering of overweight and obese population using quantile-transformed metabolic data

OBJECTIVE

Use of K-means clustering for big data technology to cluster an overweight and obese population metabolically.

METHODS

K-means clustering with the help of quantile transformation of attribute values was applied to overcome the impact of the considerable variation in the values of obesity attributes involving outliers and skewed distribution.

RESULTS

Overall, 447 subjects were categorized into six clusters; metabolically normal, mild, and severe categories. There were clearly separated metabolically normal Cluster 1 and severe Cluster 2, as well as intermediate Cluster 3, 4, and 5 that had profiles of fewer attributes with abnormal values. Cluster 3 was characteristic of sole hypertension. Cluster 3 and 4 exhibited contrasting HDL-C and LDL-C levels despite similarly elevated total cholesterol. Cluster 6 with slightly elevated triglyceride was closest to the normal group. Four- and 10-quantile-transformations yielded consistent clustering results. Compared with the original data, the quantile-transformed data produced more regular and spherical clusters and evenly distributed clusters in terms of object numbers.

CONCLUSIONS

This big data analysis strategy makes use of quantile-transformation of data to overcome the issue of outliers and the irregular distribution and applies to the analysis of other non-communicable diseases.

Use of social network analysis in the development, dissemination, implementation, and sustainability of health behavior interventions for adults: A systematic review

Interest in conceptualizing, measuring, and applying social network analysis (SNA) in public health has grown tremendously in recent years. While these studies have broadened our understanding of the role that social networks play in health, there has been less research that has investigated the application of SNA to inform health-related interventions. This systematic review aimed to capture the current applied use of SNA in the development, dissemination, implementation, and sustainability of health behavior interventions for adults. We identified 52 articles published between 2004 and 2016. A wide variety of study settings were identified, most commonly in the US context and most often related to sexual health and HIV prevention. We found that 38% of articles explicitly applied SNA to inform some aspect of interventions. Use of SNA to inform intervention development (as opposed to dissemination, implementation, or sustainability) was most common. The majority of articles represented in this review (n = 39) were quantitative studies, and 13 articles included a qualitative component. Partial networks were most represented across articles, and over 100 different networks measures were assessed. The most commonly described measures were network density, size, and degree centrality. Finally, very few articles defined SNA and not all articles using SNA were theoretically-informed. Given the nascent and heterogeneous state of the literature in this area, this is an important time for the field to coalesce on terminology, measures, and theoretical frameworks. We highlight areas for researchers to advance work on the application of SNA in the design, dissemination, implementation and sustainability of behavioral interventions.

Can big data solve a big problem? Reporting the obesity data landscape in line with the Foresight obesity system map

Background

Obesity research at a population level is multifaceted and complex. This has been characterised in the UK by the Foresight obesity systems map, identifying over 100 variables, across seven domain areas which are thought to influence energy balance, and subsequent obesity. Availability of data to consider the whole obesity system is traditionally lacking. However, in an era of big data, new possibilities are emerging. Understanding what data are available can be the first challenge, followed by an inconsistency in data reporting to enable adequate use in the obesity context. In this study we map data sources against the Foresight obesity system map domains and nodes and develop a framework to report big data for obesity research. Opportunities and challenges associated with this new data approach to whole systems obesity research are discussed.

Methods

Expert opinion from the ESRC Strategic Network for Obesity was harnessed in order to develop a data source reporting framework for obesity research. The framework was then tested on a range of data sources. In order to assess availability of data sources relevant to obesity research, a data mapping exercise against the Foresight obesity systems map domains and nodes was carried out.

Results

A reporting framework was developed to recommend the reporting of key information in line with these headings: Background; Elements; Exemplars; Content; Ownership; Aggregation; Sharing; Temporality (BEE-COAST). The new BEE-COAST framework was successfully applied to eight exemplar data sources from the UK. 80% coverage of the Foresight obesity systems map is possible using a wide range of big data sources. The remaining 20% were primarily biological measurements often captured by more traditional laboratory based research.

Conclusions

Big data offer great potential across many domains of obesity research and need to be leveraged in conjunction with traditional data for societal benefit and health promotion.