Article:Narrative Review of New Methods for Assessing Food and Energy Intake (6116053)
This page is the ScienceSource HTML version of the scholarly article described at https://www.wikidata.org/wiki/Q58789477. Its title is Narrative Review of New Methods for Assessing Food and Energy Intake and the publication date was 2018-08-10. The initial author is M. Carolina Archundia Herrera.
Fuller metadata can be found in the Wikidata link, which lists all authors, and may have detailed items for some or all of them. There is further information on the article in the footer below. This page is a reference version, and is protected against editing.
Narrative Review of New Methods for Assessing Food and Energy Intake
- M. Carolina Archundia Herrera
- Catherine B. Chan
1Department of Agriculture, Food and Nutritional Science, Alberta Diabetes Institute, University of Alberta, 6-002 Li Ka Shing Centre for Health Innovation Research, Edmonton, AB T6G 2E1, Canada; firstname.lastname@example.org
2Department of Physiology, Alberta Diabetes Institute, University of Alberta, 6-002 Li Ka Shing Centre for Health Innovation Research, Edmonton, AB T6G 2E1, Canada
Publication date (epub): 8/2018
Publication date (collection): 8/2018
Dietary self-report instruments are essential to nutritional analysis in dietetics practice and their use in research settings has facilitated numerous important discoveries related to nutrition, health and chronic diseases. An important example is obesity, for which measuring changes in energy intake is critical for assessing efficacy of dietary interventions. However, current methods, including counting calories, estimating portion size and using food labels to estimate human energy intake have considerable constraints; consequently, research on new methodologies/technologies has been encouraged to mitigate the present weaknesses. The use of technologies has prompted innovation in dietary analysis. In this review, the strengths and limitations of new approaches have been analyzed based on ease of use, practical limitations, and statistical evaluation of reliability and validity. Their utility is discussed through the lens of the 4Ms of Obesity Assessment and Management, which has been used to evaluate root causes of obesity and help select treatment options.
On a global scale, life expectancy has increased steadily for the past 35 years; however, in association with the global rise of obesity, the number of deaths from most non-communicable causes like diabetes mellitus rose by 32.1%, increasing the burden on health systems . During the past two decades, different Intensive Lifestyle Intervention programs have consistently shown that modest but clinically significant weight loss of 5% in individuals with overweight, obesity or diabetes can yield a variety of health, disease prevention and treatment benefits . Prescription of a hypocaloric diet (500–750 calories less than baseline), increased physical activity (90–175 min/week) and long term behavior change, are common techniques use in Intensive Lifestyle Intervention [,,,], which also have been described previously as individual techniques for weight control [,,].
Assessment of dietary and/or energy intake (EI) is crucial to understand the impact of clinical trials on the management of obesity and its comorbidities [,]. To date, food records (FR), food frequency questionnaires (FFQ), and 24-h recalls (24HR) are the most common methods used to assess dietary and EI during treatment and follow-up . These self-reported data methodologies have provided valuable information to use as a base to develop public health policy, comprehend and identify consumption of different food groups, understand relationship with diseases and determine eating patterns associated with weight loss, information that until recently could not be obtained in any other way .
However, a major challenge of these methods is that they rely on self-reported data. Human memory is not 100% accurate in recalling past behavior, consequently these measurements do not directly or objectively measure dietary intake or EI and do not comply with the standards of scientific methodology [,]. One issue is that the actual process of doing food records can lead individuals to change their food behavior patterns and therefore, misreport information resulting in an inaccurate report of foods, nutrients and energy consumed . Using data from The National Health and Nutrition Examination Survey (NHANES) 2003–2012, researchers analyzed the prevalence of under and over-reporting of EI, finding that in the US adult population (≥20 years) 25.1% misreported EI , results consistent with European countries where prevalence of under-reporting ranges from 20% to a high of 45% [,,,] with a predominance of obese populations under-reporting. Part of the limitation of behavior modification presented in food records can be overcome though the use of 24HR, since they can be unannounced so that the diet is not changed; however, estimation of the usual diet is weakened by recall bias (food omission or forgetfulness, erroneous estimation of portion size) . In addition to recall bias, these methods impart a substantial researcher/individual burden and high cost of administration . These methodologies used for dietary assessment have been severely criticized to the point of calling the resultant data “pseudoscientific and inadmissible in scientific research”, and what “constitutes the single greatest impediment to actual scientific progress in the fields of obesity and nutrition research” .
Thus, the accuracy of dietary assessments or modifications in dietary or EI is full of challenges and the development of new technologies to try to overcome current limitations has been encouraged . The objective of this review is to present the strengths and weaknesses of innovative new tools or methodologies that could replace, improve or complement current self-report dietary assessment instruments.
2. Materials and Methods
2.1. Search for Innovative Food and EI Assessment Tools and Methodologies
Medline, CINAHL and PsychINFO were searched for English-language articles, using the following keywords separately or in combination: diet, diet records, dietary intake, energy intake, innovate *, meals, measurement, metabolism, method, models, new, nutrition assessment, optimiz *, recent, self-report, technolog *, test reliability, test validity, trend, validation studies. The search resulted in 337 articles (Figure 1). The output was then narrowed by imposing search criteria of “2012–October 2016” and “Adults”. This search resulted in 73 articles. These article titles and abstracts were screened by one author (MCAH) to determine if they fulfilled the eligibility criteria. The articles included had to describe or validate a new method, or use new technology tools that could capture food or EI. The methodologies or tools were assessed to determine their benefits and limitations as well as their reliability or validity. Studies using text messaging or mobile phone applications that required manual introduction of information were not considered because using this type of technology imposes the same limitations as the current methods, requiring a self-reported measurement with a burdensome and impractical framework for the subject.
Of the 73 articles uncovered with this search, 17 were considered potentially eligible. These articles were cross-listed in PubMed for articles related to the topic, which identified 8 other articles. Reference lists of relevant articles were also hand-searched but no other relevant articles were found. For these 25 articles, review of the full text was used to identify those meeting the criteria (n = 11).
A relative evaluation of the innovative technology tools and methodologies was carried out. They were assessed to ensure that the main weaknesses of present methodologies: recall biases, measurement discrepancies, lack of scientific rigor, were being acknowledged.
When developing tools to collect dietary information, specific statistical methods must be used to evaluate their reliability and validity in order to test the accuracy of the method and avoid bias . Thus, utilization of these recommended statistical methods to assess the different tools and methodologies was noted when drafting this manuscript. Reliability refers to “the consistency of a measuring instrument” , in different situations; inter-rater, test-retest, inter-method and internal consistency. Validity refers to “how close the tool can measure the actual (true) value”; in this case, a measure of true EI when compared to the gold standard .
Benefits and limitations: When describing benefits and limitations of each tool/methodology, the focus was on the following criteria: Easy to administer—Referring to reducing participant burden. Current methods rely on information reported by the subjects recalling what they ate for the past week/month/year, or keeping a diary for various days. This decreases the quality of the reports, and the process itself can make subjects change their eating habits . Easy to score—Observed and weighed-food records , doubly labelled water (DLW) , and FFQ , are some of the methods currently used to determine/estimate EI and/or eating behavior. These methods are expensive and time-consuming, making them less feasible to use and hard to score. New methodologies should minimize practitioners’ or researchers’ burden and expense. Capture change over more than one day—Because assessment of day-to-day variability in food intake is an important limitation of current methodologies. New methodology should overcome these limitations and be able to capture fluctuations in habitual energy and nutrient intake on free-living subjects.
Table 1 summarizes the studies that were included in the review. Five described different types of monitors and sensors; five described camera-scan-sensor-based technologies; and one described a mathematical method. Details of the statistical methods used to assess validity and reliability are noted in Table 2.
3.1. Food/Energy Intake Monitoring Devices and Tools
Use of body sensors as a direct measurement of human eating behavior is quite recent. Body monitors and sensors have been developed with the hope of improving and facilitating measurement of daily food and EI [,].
3.1.1. Automated Wrist Motion Tracking
The Automated Wrist Motion Tracking, also called a “bite counter” is worn like a watch and automatically tracks wrist motion for monitoring eating in humans . Reliability was tested in both controlled meal and semi-controlled settings. The sensitivity was >85% in both settings. Bites measured by the device were >80% detected compared with bites counted by direct observation. The equations used to measure sensitivity and performance are reported in Table 2 . A third experiment in free-living situations was performed to examine the correlation between bites detected and EI, with r = 0.6. This experiment was only exploratory and was done to seek any possible relationships between these factors for further research .
Use of this device resulted in improved accuracy of measuring EI in free-living situations compared with 24HR and FFQ, which typically under-report EI in men by 16–20% and 31–36%, and in women by 16–20% and 34–38% respectively . Participant burden was minimal because the user only needed to turn it on and off before eating, and thereafter, bites were registered automatically by the device; thus, researcher and administrative costs are ameliorated since no food weight or labour-intensive laboratory techniques are needed [,,]. Forgetting to use the device, accuracy in different social settings and loss of data when both hands are used to eat are present limitations that the bite counter tool needs to address. Importantly this device’s main benefit is its use as a food intake-monitoring and ingestive behavior tracking system in a real-world setting to improve users, researches and HCP understanding of food intake behaviors. Furthermore, lessen the burden of manual measurements; however, no input regarding the type or quality of the food consumed is tracked.
3.1.2. The Bite-Based Model of Kilocalorie Intake
The bite-counter described above was used for the development of a kilocalorie per bite equation (using bite counts, individual demographic and physical characteristics) that allowed EI to be estimated. The relationship obtained was estimated as kilocalories per bite = −0.128 age + 6.167 sex (female = 0) + 0.034 height + 0.035 weight − 12.012 WHR + 22.294; where WHR = waist-to-hip ratio . The feasibility of using the formula was then systematically evaluated. Two trials were run using a train and test paradigm, in which the training group was used to develop the model and the test group was used to determine reliability of the regression model.
When comparing the reliability of the formula-predicted EI values to the staff-observed values within the test group the Pearson correlation was r = 0.374. For the reliability of the model between training and test groups, the difference in r2 (which is called the shrinkage value) was 1.4% . To assess validity, researchers assessed participants’ estimation error of EI compared with the equation. The bite-based equation method was more effective at estimating EI than the best human estimation .
The bite counter along with the bite-based method formula can provide individuals with a EI estimation that is more accurate than an individual’s estimation even when EI information is available, which potentially could help improve their adherence to recommended dietary changes. The bite counter also has the benefit of being a non-invasive device, which allows tracking of free-living situations for research and also has the potential to improve the understanding of food ingestion patterns including snacking, night eating, and weekend overeating, as pointed out by Fontana 2014  as one of the benefits of food intake monitoring devices and tools. However, reliability was relatively low and internal and external validity of the method needs to be further elucidated. This tool is for monitoring EI purposes only and does not provide information or feedback on diet quality.
3.1.3. The Automatic Ingestion Monitor (AIM)
The Automatic Ingestion Monitor (AIM) integrated hand gestures, jaw motions and accelerometer sensors to detect food intake in free-living individuals . It was designed for objective 24-h monitoring of food intake in free-living conditions without depending on any input from the subjects. The monitor was 90% accurate in its ability to detect specific food intake epochs in free-living individuals compared with self-reported signal (push-button) indicating food intake events, and self-reported food journals .
When developing and validating this device, the data were obtained from monitoring free-living situations that included a wide variety of foods and activities, increasing its feasibility for everyday use and research purposes. Its use could provide insight into overall eating behavior patterns where participants burden is minimal. Nonetheless, the use of self-report as the gold-standard method, rather than direct observation, prompts caution regarding reliability. Furthermore, subject compliance with and acceptability of wearing the AIM needs to be established . Insight obtain from this and previous studies  encourage further research to build mathematical models to obtain estimated EI using individualized models on counts of chews and swallows (CCS) .
3.1.4. Intelligent Food-Intake Monitor
The intelligent food-intake monitor integrates multi-sensor monitors to track chewing speed, and images of the type and amount of food consumed, giving an overall understanding of eating behavior characteristics . The tool was tested for its ability to correctly detect the proportion of food consumed in real life scenarios but results were not reported .
The development of the device took into consideration the general process and pattern of food-intake activities to directly target their process (food ingestion, chewing and swallowing). The experiments were conducted in a real-world setting to increase the feasibility of being used in such settings. Valuable information involving eating behavior can be obtained from the use of this device because it doesn´t assume that the food on the plate is consumed, thus providing a more reliable measure than capturing images of food alone to assess food consumption, thanks to the integration of chewing and swallowing detection in the process. Further research needs to be conducted to increase participants’ comfort levels when using the device to ensure compliance with its use for longer periods. Even though a high level of correlation is reported between ground truth and auditory and vision predictors, no r-values were given and no strong statistical bases were presented. Participant characteristics were not supplied , leaving to speculation the age range that could benefit from using this monitor, and whether it would be feasibility to use in older adults, youth and children.
3.2. Camera-Scan-Sensor Based Technologies or Food/Energy Intake Assessment Tools
Sixty-four percent of the American population own a smartphone, a 35% increase since 2011 . Since the use of smartphones is steadily increasing in daily life, mobile phone camera-scan-sensors are being proposed to contribute novel approaches to the measurement of food and EI.
3.2.1. DP + R
Ptomey et al.  developed and evaluated a pre-post meal photographic method for assessing EI in overweight and obese individuals in a cafeteria setting. Foods consumed outside this setting were assessed by recall methods.
Nutrition research staff underwent rigorous training for estimating portion size and EI from pre- and post-meal digital photographs and dietary recalls, with inter-rater reliability >95%. DP + R procedure includes taking notes and delineating standard measurements as guidelines for the portion size assessment . The DP + R during ad libitum eating in a cafeteria was compared to measurement of total daily energy expenditure assessed by doubly labelled water (TDEEDLW) method  with no significant differences found; thus, the method was considered valid.
DP + R method is a reliable and validated method for estimating EI in overweight or obese participants in a cafeteria setting. The main advantage over a food record/recall alone is verification of the written record by the photograph. This method was judged to provide an acceptable level of burden for both participant and research team when compared to previous procedures but a considerable burden is still present for the researcher because of the need to enter nutritional information into a database to quantitate EI . The capacity of DP + R to capture change over time is limited since the procedures are done in cafeteria settings, therefore when the subjects stop attending, the change will not be captured. However, the authors point out the possibility of modifying the DP + R method to use in conjunction with smartphone photos to make the method portable .
3.2.2. Remote Food Photography Method (RFPM)
Participants send images taken on their smartphone wirelessly  to a Food Photography Application© , which is linked to the Food and Nutrient Database for Dietary Studies 3.0 . Trained raters use the application to oversee the semi-automated process of food and nutrient intake estimation . In this trial, some participants received prompts to use the Application customized to their specific meal times, or generic prompts in the morning, at noon and in the late afternoon.
Analyses were run to evaluate any significant differences between the RFPM and DLW estimation of EI, and if they were influenced by the EI consumed; no significant differences were found when participants received customized reminder messages but device reliability was decreased when participants received generic prompts . The RFPM and DLW were used to measure EI in free-living individuals during a 6-day period. The error between methods (EI estimated with the RFPM minus EI measured with DLW) was calculated and was smaller in the participants receiving customized prompts .
The underestimation of EI by RFPM improves drastically compared to self-report methods, particularly when accompanied by customized prompts, allowing monitoring of habitual EI in free-living individuals. The method also offers the opportunity to detect missing data (due to technical problems or no compliance) promptly, and take pertinent action (contact participant) to improve data quality and compliance, thereby reducing recall bias . The ability of RFPM to provide users feedback about their behavior is another benefit worth mentioning. In general, the user burden is keep to a minimal and 82% of users rated overall satisfaction 5 or higher (based on a six-point scale) . However, since the method is only semi-automated, it remains expensive to analyze. The RFPM has also been used to estimate EI in children in both research and free-living settings [,,].
3.2.3. Real-Time Food Recognition System
The user points the smartphone camera at the food plate for the food recognition process. After selection of the food from a database and indication of its approximate volume, the calorie and nutrition values are displayed . The real-time recognition of foods was approximately 80% correct.
This system utilizes a real-time image recognition system, and the processing time only takes 0.065 s once the user enters the input. A fully automated interface with a food database completes its system. Evaluation of its usability was carried out, where adjustment of the bounding boxes on the different food items wasn’t as positively rated (2.4 out of 5) as for the item recognition itself, which was done automatically without additional user input, obtaining an average score of 4.2 out of 5 . However, this tool hasn’t been validated and has a limited number of food categories and it does not specify the database used for the nutrition information.
A “snap-shot” (photograph) of participants’ plate is captured. The analytical system is based on predefined EI and nutritional density for each food category. Depth images are used to estimate the portion size of the food and the EI and nutritional content are displayed on the user’s screen in ~4 s . A classification accuracy (the percentage of the test images of each category correctly classified) of 85% was obtained for 15 different food categories.
Snap-n-Eat presents a food recognition system for which users only need to take a snapshot of their food in order for the system to estimate its EI and nutritional content allowing participants to track their daily food intakes helping to understand their eating habits in a cost and time effective manner. However, in order to be a feasible tool, a scale-up to hundreds of food items and a validation process is needed .
The user photographs their food from at least two angles. The food items are segmented and recognised and their carbohydrate content is estimated based on the nutritional information of the USDA Nutrient Database for Standard Reference . GoCARB’s portion sizing and individual food item recognition accuracy ratings were 75% and 85%, respectively . To validate the device, adult participants with type 1 diabetes were asked to calculate the carbohydrate content of the meals by themselves and subsequently with the help of the GoCARB. The error using GoCARB error was approximately half of that without any aid .
The application is overall better than participants at estimating carbohydrate content of meals. In the GoCARB app the carbohydrate content estimation is done automatically so the burden on researchers and participants is minimal; thus, 90% (17/19) qualify the tool as easy to use and would like to use the application on a regular basis. These measurements were done in a clinical setting that may not represent real-life situations where the meals may have more complex composition than the test meals. The overall nutrient content of the meal is not analyzed, and for individuals with diabetes it is important to consider the influence of the overall meal in determining their postprandial glycemia .
3.3. Mathematical Algorithm
A totally different, novel approach to assess EI is through mathematical algorithms.
Sanghvi’s group  validated a mathematical formula originating from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) , to measure long-term changes in free-living EI of humans by using repeated DLW/DXA measurements collected over 2 years in 140 free-living subjects from the Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy (CALERIE) study . The formula inputs were baseline demographic (age, sex, height) and repeated body weight data, which was used to obtain the change in body weight over time and the rolling average.
Measured body weight and EI changes for the participants were documented over 2 years at 4 different time intervals. During the course of the study, the test-retest reliability was obtained by comparing the gold standard with the mathematical model, differing by only 40 kcal/day . The change in EI values calculated by the mathematical method was compared (paired, 2-sided t-test) to the gold standards DLW/DXA and found to be similar .
In order for the formula to measure long-term changes in free-living EI, easily acquired initial information regarding age, sex, height and physical activity are required. However, baseline DLW measurements are also needed to establish energy requirements if one wishes to know absolute EI as well as changes in EI over time, limiting its use to researchers with the ability to obtain this parameter . If all the mathematical parameters are available, the formula is an easy-to-score tool that captures changes over more than one day; however, the model might require adjustments for use in children or older adults. In addition, specific nutrient/food intake information is not known, therefore without co-administration of a diet record or FFQ it would not be possible to obtain this information . Important limitations must be considered. The study was conducted on normal weight individuals and validation in individuals with obesity was not demonstrated even though the authors were confident the model could be used on this population because the model was built to measure changes in metabolism and body composition.
From a research perspective, the first and foremost goal of evaluating food and EI is to be able to increase our understanding of diet-disease associations. Validated and reliable measures of food and EI are crucial to understand their relationship with health, especially with the overwhelming increase in obesity prevalence . Individuals with obesity present different problems ranging from the physiological to the psychological aspects, which represent barriers to their treatment. The 4Ms of Obesity Assessment and Management (Mental, Mechanical, Metabolic and Monetary) has been proposed as a framework to help identify the root cause and help obesity treatment .
The methodologies/tools presented in this review have the potential to aid in the understanding and treatment of obesity within this framework. This review identified 3 main new modalities for estimating food and EI. These include devices that monitor intake through sensors that detect movement of the arm and/or jaw, counts of chews and swallows, smartphone-based photographic methods linked to food databases and a mathematical formula.
In order to come to a consensus of which methodology/technology would be the most highly recommended it is important not to lose sight of why EI is being assessed or monitored. The overall objective should guide opting for one or the other.
In the context of the 4Ms of Obesity Assessment and Management, if the individual being treated is believed to have psychological (Mental) issues influencing their eating behavior, then the main objective is to understand their eating behaviors or food intake patterns in order to detect and/or modify eating habits. Food intake-monitoring devices and tools (Bite Counter, AIM, Intelligent food-intake monitor) would be recommended in this context because they could provide useful insight regarding food intake behaviors (e.g., timing and size of meals). In general, food intake-monitoring devices and tools can count the number of bites an individual takes, track the approximate EI and monitor episodes of food intake. Several benefits to the understanding of food intake behaviors may accrue from these methods. These devices could fill a gap in providing timely monitoring and feedback to individuals wishing to change eating habits, similar to the way the use of pedometers and/or accelerometers has been validated to promote and assess physical activity , by establishing and monitoring personal goals achievement, a behavior that according to Social Cognitive Theory, is an effective behavior change strategy  aiding behavioral change. Therefore, these tools could be used for monitoring, controlling and correcting eating behaviors and portion size in obese or overweight individuals as well as for chronic disease management. However, their effectiveness in eliciting behavior change has yet to be documented. Future work includes the possible addition of a vibrotactile alarm, similar to the technology used on intelligent watches or pedometers so that subjects can self-adjust their eating behavior based on the estimated EI per bite . Moreover, the commercial cost of these devices has not been established since they are still on the development phase and have not gone further to establish a market cost. Further, wearing some devices may be more acceptable to participants than others.
On the other hand, if the aspect of obesity treatment is within the Metabolic category of the 4Ms, as in the case of individuals with T2D or hypertension, then the intent would shift the focus to understanding specific macro/micronutrient intakes (sugars, salt, fats). Similarly, within the Mechanical category (such as osteoarthritis), weight loss could be desired to reduce pain. For both approaches, camera-scan-sensor (Snap-n-Eat, GoCARB) could be useful. RFPM could be applied in a hospital setting where monitoring individuals’ nutrition intake is essential but difficult to do on a routine basis. Registered dietitians and nurses could use this tool to oversee adequate food intake essential for hospitalized individuals’ wellbeing. If the overall objective is a focus on measures of long-term changes in EI in free-living individuals undergoing a research or lifestyle intervention, the mathematical method would highly be recommended since its accuracy lies within 40 kcal/day of mean difference with the gold standard, as long as the initial DLW measurement is possible to obtain, which could be a potential limitation.
Overall, the studies included in this review presented new devices designed to improve how EI is measured, analyzed and registered. However, the devices and methods have usually undergone pilot testing in small numbers of participants and various limitations elicit caution. Food intake-monitoring tools have limited ability to assess day-to-day variability in food intake [,,]. They do not take into account the type of food consumed, its EI density nor its consistency; therefore, no information about the macro/micronutrient is obtained, resulting in a inability to capture change in type of food or nutritional intake over time. As mentioned previously, a current limitation with present methodologies used to assess food intake or EI is individual reactivity causing changes in food behavior patterns, thereby resulting in inaccurate reporting. None of the present studies addressed these issues, therefore the question arises: could bias play a role in the use of these devices? That is, would peoples’ consumption of food intake be modified by simply wearing these tools? And if so, what would be the differences compared with current methodologies? Certainly, more accurate data of consumption patterns seems possible, but to date none of the devices has gone beyond pilot testing nor addressing potential bias. To our knowledge, the application of these methodologies to clinical settings or outside of the original developers’ laboratories has not been reported.
Regarding smartphone-based apps, additional limitations applying to one or more include participants forgetting to take the photographs, or not having the smartphone with them . In general, two major limitations need to be addressed with camera-scan-sensor methods. First, they cannot quantify all food ingredients or beverages. These tools only work with the food items in the database of each individual tool, and their validity is also dependent on the food nutrient value on which the databases are built. However, with current food record/recall databases, there are acknowledged differences between what a person consumes and what the database contains ; even the Canadian Nutrient File or the USDA database cannot keep up with constantly evolving food possibilities. The use of these technologies is not advanced enough to correctly and accurately estimate 100% of food intake since the best achieved accuracy was 85% based on a small number of foods . Second, they cannot judge quality since a photograph doesn’t convey information about ingredients that are hidden or blended . Nevertheless, these methods show improvement in estimating, on average, the nutrient content of meals more accurately, easier and faster than individuals’ self-report measures (24HR, FFQ, etc.) but caution must be taken when using and analyzing these methods. Bearing in mind the strong link between food intake and health, continuing to document the improved validity and reliability of the food item recognition and nutritional information provided by these tools would undoubtedly lead to better outcome measurement in the fields of obesity and nutrition. However, the feasibility creating comprehensive databases for food recognition is problematic in an environment of incessantly increasing food possibilities. On the other hand, the ubiquity of smartphone ownership means that affordability and acceptability are of less concern with the main investment being the data processing.
In conclusion, these innovative dietary assessment tools are able to record food/energy intake more accurately than participants’ estimates and are an improvement on important weaknesses of conventional methods (paper-based records/recalls), particularly regarding the burden of recording by participants and collecting/administering and evaluating/scoring the information by researchers. However, caution is needed when using them since they are still being refined. Future work should look at combining body monitor sensors and camera-scan-sensors to work together in order to counter their strengths and weaknesses. This work should eventually progress outside of research settings and promote the collaboration of dietitians with engineers to co-develop the design, development, evaluation and implementation of these new tools, since this would likely increase their effectiveness, acceptability and validity. Lastly, this research field should take into consideration changing formats of national nutrition recommendations, such as the 2014 Brazilian dietary guidelines, 2015 Dietary Guidelines for Americans and the American Heart Association, which are shifting the focus from single nutrients or kilocalorie counting into healthy eating patterns [,,]. Therefore, future development should aim at being able to detect overall eating patterns.
M.C.A.H. was supported by the National Council of Science and Technology of Mexico.
- H. WangM. NaghaviC. AllenR.M. BarberZ.A. BhuttaA. CarterD.C. CaseyF.J. CharlsonA.Z. ChenM.M. CoatesGlobal, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for 249 causes of death, 1980–2015: A systematic analysis for the global burden of disease study 2015Lancet20163881459154410.1016/S0140-6736(16)31012-127733281
- D.A. WilliamsonFifty years of behavioral/lifestyle interventions for overweight and obesity: Where have we been and where are we going?Obesity2017251867187510.1002/oby.2191428944593
- S.K. DasS.B. RobertsM.V. BhapkarD.T. VillarealL. FontanaC.K. MartinS.B. RacetteP.J. FussW.E. KrausW.W. WongBody-composition changes in the comprehensive assessment of long-term effects of reducing intake of energy (CALERIE)-2 study: A 2-y randomized controlled trial of calorie restriction in nonobese humansAm. J. Clin. Nutr.201710591392710.3945/ajcn.116.13723228228420
- The Diabetes Prevention Program (DPP) Research GroupThe Diabetes Prevention Program (DPP): Description of lifestyle interventionDiabetes Care2002252165217112453955
- The Look AHEAD Research GroupEight-year weight losses with an intensive lifestyle intervention: The Look AHEAD studyObesity20142251324307184
- W. MaT. HuangY. ZhengM. WangG.A. BrayF.M. SacksL. QiWeight-loss diets, adiponectin, and changes in cardiometabolic risk in the 2-year pounds lost trialClin. Endocrinol. Metab.20161012415242210.1210/jc.2016-120727055193
- B.K. SharmaM. PatilA. SatyanarayanaNegative regulators of brown adipose tissue (BAT)-mediated thermogenesisJ. Cell Physiol.20142291901190710.1002/jcp.2466424809334
- D.H. RyanM.D. JensenWhat the new obesity guidelines will tell usCurr. Opin. Endocrinol. Diabetes Obes.20132042943310.1097/01.med.0000432612.80860.3423974771
- S.N. BlairPhysical inactivity: The biggest public health problem of the 21st centuryBr. J. Sports Med.2009431219136507
- A.F. SubarL.S. FreedmanJ.A. ToozeS.I. KirkpatrickC. BousheyM.L. NeuhouserF.E. ThompsonN. PotischmanP.M. GuentherV. TarasukAddressing current criticism regarding the value of self-report dietary dataJ. Nutr.20151452639264510.3945/jn.115.21963426468491
- R.K. JohnsonDietary intake—How do we measure what people are really eating?Obes. Res.20021063S68S10.1038/oby.2002.19212446861
- E. ArcherG. PavelaC.J. LavieThe inadmissibility of what we eat in America and NHANES dietary data in nutrition and obesity research and the scientific formulation of national dietary guidelinesMayo Clin. Proc.20159091192610.1016/j.mayocp.2015.04.00926071068
- N.V. DhurandharD. SchoellerA.W. BrownS.B. HeymsfieldD. ThomasT.I. SorensenJ.R. SpeakmanM. JeansonneD.B. AllisonEnergy Balance Measurement Working Group. Energy balance measurement: When something is not better than nothingInt. J. Obes.2015391109111310.1038/ijo.2014.19925394308
- J.P. IoannidisImplausible results in human nutrition researchBMJ2013347f669810.1136/bmj.f669824231028
- K. MurakamiM.B. LivingstonePrevalence and characteristics of misreporting of energy intake in US adults: NHANES 2003–2012J. Nutr.20151141294130310.1017/S000711451500270626299892
- I. Berta VanrullenJ.L. VolatierA. BertautA. DufourJ. DallongevilleCharacteristics of energy intake under-reporting in French adultsJ. Nutr.20141111292130210.1017/S000711451300375924330816
- L. GemmingY. JiangB. SwinburnJ. UtterC.N. MhurchuUnder-reporting remains a key limitation of self-reported dietary intake: An analysis of the 2008/09 New Zealand adult nutrition surveyJ. Clin. Nutr.20146825926410.1038/ejcn.2013.24224300904
- K. MurakamiT.A. McCaffreyM.B. LivingstoneAssociations of dietary glycaemic index and glycaemic load with food and nutrient intake and general and central obesity in British adultsJ. Nutr.20131102047205710.1017/S000711451300141423656860
- L. JohanssonK. SolvollG.E. BjorneboeC.A. DrevonUnder- and overreporting of energy intake related to weight status and lifestyle in a nationwide sampleAm. J. Clin. Nutr.19986826627410.1093/ajcn/68.2.2669701182
- P.P. BasiotisS.O. WelshF.J. CroninJ.L. KelsayW. MertzNumber of days of food intake records required to estimate individual and group nutrient intakes with defined confidenceJ. Nutr.19871171638164110.1093/jn/117.9.16383655942
- F.E. ThompsonA.F. SubarC.M. LoriaJ.L. ReedyT. BaranowskiNeed for technological innovation in dietary assessmentJ. Am. Diet. Assoc.2010110485110.1016/j.jada.2009.10.00820102826
- V. BountzioukaD.B. PanagiotakosStatistical methods used for the evaluation of reliability and validity of nutrition assessment tools used in medical researchCurr. Pharm. Des.2010163770377510.2174/13816121079445510221128895
- S.M. RebroR.E. PattersonA.R. KristalC.L. CheneyThe effect of keeping food records on eating patternsJ. Am. Diet. Assoc.1998981163116510.1016/S0002-8223(98)00269-79787724
- M.E. HiseD.K. SullivanD.J. JacobsenS.L. JohnsonJ.E. DonnellyValidation of energy intake measurements determined from observer-recorded food records and recall methods compared with the doubly labeled water method in overweight and obese individualsAm. J. Clin. Nutr.20027526326710.1093/ajcn/75.2.26311815316
- D.A. SchoellerMeasurement of energy expenditure in free-living humans by using doubly labeled waterJ. Nutr.19881181278128910.1093/jn/118.11.12783142975
- O. AmftG. TrosterRecognition of dietary activity events using on-body sensorsArtif. Intell. Med.20084212113610.1016/j.artmed.2007.11.00718242066
- E. SazonovS. SchuckersP. Lopez-MeyerO. MakeyevN. SazonovaE.L. MelansonM. NeumanNon-invasive monitoring of chewing and swallowing for objective quantification of ingestive behaviorPhysiol. Meas.20082952554110.1088/0967-3334/29/5/00118427161
- Y. DongA. HooverJ. SciscoE. MuthA new method for measuring meal intake in humans via automated wrist motion trackingAppl. Psychophysiol. Biofeedback20123720521510.1007/s10484-012-9194-122488204
- A.F. SubarV. KipnisR.P. TroianoD. MidthuneD.A. SchoellerS. BinghamC.O. SharbaughJ. TrabulsiS. RunswickR. Ballard-BarbashUsing intake biomarkers to evaluate the extent of dietary misreporting in a large sample of adults: The open studyAm. J. Epidemiol.200315811310.1093/aje/kwg09212835280
- J.N. SalleyA.W. HooverM.L. WilsonE.R. MuthComparison between human and bite-based methods of estimating caloric intakeJ. Acad. Nutr. Diet.20161161568157710.1016/j.jand.2016.03.00727085871
- J.M. FontanaM. FarooqE. SazonovAutomatic ingestion monitor: A novel wearable device for monitoring of ingestive behaviorIEEE Trans. Biomed. Eng.2014611772177910.1109/TBME.2014.230677324845288
- J.M. FontanaJ.A. HigginsS.C. SchuckersF. BellisleZ. PanE.L. MelansonM.R. NeumanE. SazonovEnergy intake estimation from counts of chews and swallowsAppetite201585142110.1016/j.appet.2014.11.00325447016
- J. LiuE. JohnsL. AtallahC. PettittB. LoG. FrostG.-Z. YangAn intelligent food-intake monitoring system using wearable sensors2012 Ninth International Conference on Wearable and Implantable Body Sensor NetworksIEEELondon, UK2012154160
- L.T. PtomeyE.A. WillisJ.J. HonasM.S. MayoR.A. WashburnS.D. HerrmannD.K. SullivanJ.E. DonnellyValidity of energy intake estimated by digital photography plus recall in overweight and obese young adultsJ. Acad. Nutr. Diet.20151151392139910.1016/j.jand.2015.05.00626122282
- C.K. MartinJ.B. CorreaH. HanH.R. AllenJ.C. RoodC.M. ChampagneB.K. GunturkG.A. BrayValidity of the remote food photography method (RFPM) for estimating energy and nutrient intake in near real-timeObesity20122089189910.1038/oby.2011.34422134199
- Y. KawanoK. YanaiFoodcam: A real-time food recognition system on a smartphoneMultimed. Tools Appl.2015745263528710.1007/s11042-014-2000-8
- W. ZhangQ. YuB. SiddiquieA. DivakaranH. Sawhney“Snap-n-eat”: Food recognition and nutrition estimation on a smartphoneJ. Diabetes Sci. Technol.2015952553310.1177/193229681558222225901024
- D. RhynerH. LoherJ. DehaisM. AnthimopoulosS. ShevchikR.H. BotweyD. DukeC. StettlerP. DiemS. MougiakakouCarbohydrate estimation by a mobile phone-based system versus self-estimations of individuals with type 1 diabetes mellitus: A comparative studyJ. Med. Int. Res.201618e10110.2196/jmir.556727170498
- A. SanghviL.M. RedmanC.K. MartinE. RavussinK.D. HallValidation of an inexpensive and accurate mathematical method to measure long-term changes in free-living energy intakeAm. J. Clin. Nutr.201510235335810.3945/ajcn.115.11107026040640
- E.S. SazonovS.A.C. SchuckersP. Lopez-MeyerO. MakeyevE.L. MelansonM.R. NeumanJ.O. HillToward Objective Monitoring of Ingestive Behavior in Free-living PopulationObesity2009171971197510.1038/oby.2009.15319444225
- Pew Research CenterThe Smartphone DifferenceAvailable online: http://www.pewinternet.org/2015/04/_1/us-smartphone-use-in-2015/(accessed on 1 April 2015)
- C.K. MartinT. NicklasB. GunturkJ.B. CorreaH.R. AllenC. ChampagneMeasuring food intake with digital photographyJ. Hum. Nutr. Diet.201427728110.1111/jhn.1201423848588
- USDA Food and Nutrient Database for Dietary Studies, version 3.0 [internet], Beltsville (MD): USDA, Agricultural Research Service, Food Surveys Research Group2008Available online: http://www.ars.usda.gov/Services/docs.htm?docid=17031(accessed on 1 April 2015)
- C.K. MartinH. HanS.M. CoulonH.R. AllenC.M. ChampagneS.D. AntonA novel method to remotely measure food intake of free-living individuals in real time: The remote food photography methodJ. Nutr.200910144645610.1017/S000711450802743818616837
- K.D. HallC.C. ChowEstimating changes in free-living energy intake and its confidence intervalAm. J. Clin. Nutr.201194667410.3945/ajcn.111.01439921562087
- J. RochonC.W. BalesE. RavussinL. RedmanJ.O. HolloszyS.B. RacetteS.B. RobertsS.K. DasS. RomashkanK.M. GalanDesign and conduct of the CALERIE Study: Comprehensive assessment of the long-term effects of reducing intake of energyJ. Gerontol. Ser. A201066A9710810.1093/gerona/glq16820923909
- A.M.M. SharmaA mnemonic for assessing obesity. Obesity reviewsOff. J. Int. Assoc. Study Obes.20101180880910.1111/j.1467-789X.2010.00766.x21182728
- H. Van RemoortelS. GiavedoniY. RasteC. BurtinZ. LouvarisE. Gimeno-SantosD. LangerA. GlendenningN.S. HopkinsonI. VogiatzisValidity of activity monitors in health and chronic disease: A systematic reviewInt. J. Behav. Nutr. Phys. Act.201298410.1186/1479-5868-9-8422776399
- A. BanduraSocial Foundations of Thought and Action: A Social Cognitive TheoryPrentice-Hall, Inc.Englewood Cliffs, NJ, USA1986617
- C.A. MonteiroG. CannonJ.C. MoubaracA.P. MartinsC.A. MartinsJ. GarzilloD.S. CanellaL.G. BaraldiM. BarciotteM.L. LouzadaDietary guidelines to nourish humanity and the planet in the twenty-first century. A blueprint from brazilPublic Health Nutr.2015182311232210.1017/S136898001500216526205679
- B.E. MillenS. AbramsL. Adams-CampbellC.A.M. AndersonJ.T. BrennaW.W. CampbellS. ClintonF. HuM. NelsonM.L. NeuhouserThe 2015 dietary guidelines advisory committee scientific report: Development and major conclusionsAdv. Nutr.2016743844410.3945/an.116.01212027184271
- D.M. Lloyd-JonesY. HongD. LabartheD. MozaffarianL.J. AppelL. Van HornK. GreenlundS. DanielsG. NicholG.F. TomaselliDefining and setting national goals for cardiovascular health promotion and disease reduction: The american heart association’s strategic impact goal through 2020 and beyondCirculation201012158661310.1161/CIRCULATIONAHA.109.19270320089546