In the United Kingdom (UK), a national survey gave the estimates that 72% of the population owned a smartphone , and 80% of adults used the Internet on a daily or almost daily basis in the year 2017 . Smartphone applications, computers and the Internet are widely used in several aspects of modern life, including health communication; widespread use of these technologies also provides opportunities for dietary assessment.
Technologies used for dietary assessment include a computer, the Internet, telecommunication and imaging technologies. In most cases, these technologies have the same aim as the paper-based subjective methods of dietary assessment but aim to improve the accuracy of the assessment and compliance of the participants. Overall, a preference of technology-assisted approaches over traditional methods has been shown among both adolescents and adults [48, 28]. Computer displays of portion sizes can be used for dietary interviews to aid accurate estimation of the amount of food and drink consumed. However, using these technology-based aids has a limited impact on accuracy as they still rely on participants’ judgement and recall of portion sizes . While this example highlights needs for careful appraisal of available technologies and their applications, modern technologies with automated data capturing and coding systems have nonetheless the potential to overcome the inherent limitations of traditional methods such as:
Dietary assessment with innovative technologies is based on an electronic recording of foods and drinks consumed. For example, participants can report foods from a photo list of foods in a database or they can take photographs of foods, which later can be analysed by researchers or automatic algorithms to derive dietary intakes. On the whole, technology-assisted methods can be broadly categorised, based on the type of technology used, as shown in Figure D.2.7. At the broadest level, they can be distinguished as 1) computerised/smartphone app versions of traditional subjective pen-and-paper methods, and 2) computerised/smartphone app which captures additional raw data over and above traditional methods. The sections below follow this structure.
Figure D.2.7 Classification of technologies used for dietary assessment.
1. Computerised/smartphone app versions of traditional pen-and-paper methods
1.1 Dietary assessment with ‘static’ computerised technologies
Computerised dietary assessment technologies can be developed as self-administered dietary assessment instruments. Often with the Internet connected to a server established by the researchers, computer-based technologies collect similar data to commonly used paper-based dietary assessment tools (e.g. food frequency questionnaire and 24-hour dietary recall). Yet, computer-based assessment tools have advantages over manual tools. For example, participants can be guided to accurately report foods with a specific brand through a visual list of a large number of food items, including packages and appearances, while this is practically not possible in traditional assessment. This method is supposed to let participants recall their food consumption in different occasions in details in a standardised manner. More details regarding food photographs as aids to the portion size estimation are available here.
Interviewer-based dietary assessment can also be aided by static computerised technologies. An interview needs to be well structured to help a participant report dietary consumption including types of foods and beverages, their quantity, brands, time, and occasions. In addition to visual aids, structural guide for an interviewer to follow reduces the chance of omitting certain dietary queries, such as those about dietary supplements or snacking. In a clinical setting, which can be coupled with dietary intervention, a computer-assisted interviewer-based dietary assessment is of a strong option.
Figure D.2.8 Examples of self-administered web-based technologies.
1.2 Dietary assessment with mobile application technologies
These methods have similar approaches to ‘static’ computerised dietary assessment technologies listed above, with the additional feature of portability. However, the most prevalent smartphone apps are those which use digital food photography to assist the dietary assessment as advancement to the traditional methods (see section 2 below).
Some apps include a food and nutrient database, which is linked with barcodes of the foods that are purchased, so the users can scan the food labels with a barcode scanner and the nutrients can be derived .
2 Computerised/smartphone app which capture additional raw data
2.1 Dietary assessment with non-automated digital food photography
These methods are based on capturing images before and after eating episodes to provide primary records of dietary intake instead of manual recording. The photograph must be taken manually. These methods require processes to digitalise photo images into dietary data. These methods typically make use of the advanced technologies of smartphones such as wireless communication, built-in cameras, global position system, portable design and external devices connectivity such as Bluetooth .
For better estimation of colour and portion size by experts, participants use a fiducial marker, a reference item such as a pen or colour checkerboard placed within a camera frame while capturing images (Figure D.2.9).
Some methods using food photos have been shown to be reliable and accurate measure of food intakes, both among adults [17, 12, 43] and children [33, 30]. Previous studies also showed higher preferences (between 91 to 100%) of these methods compared to pen-and-paper methods among participants .
Procedures for data collection with these methods are as follows:
Figure D.2.9 Food images with the use of fiducial markers.
My Meal Mate (MMM) (Figure D.2.11) .
Pattern - Oriented Nutrition Diary (POND) 
Weight Management Mentor (WMM) 
Dietary Intake Monitoring Application (DIMA) 
Figure D.2.10 Remote Food Photography Method (RFPM).
Recaller app 
Nutricam Dietary Assessment Method ;
Figure D.2.11 Screen capture of the food diary entry page of My Meal Mate.
The advanced versions of these methods are based on automated nutrient derivation. These methods are based on automated image segmentation and analyses from food images captured by participants and intended to reduce manual image analysis by dietary assessment expert (Figure D.2.12). Based on image segmentation, colour and texture features are extracted which lead to food classification. Accuracy of segmentation depends on the number of foods sent at a time and type of food. Accuracy of food classification depends on the corresponding number of images available in the reference database.
Figure D.2.12 DietCam system architecture.
Mobile device food record (mdFR)
This method was previously based on manual image analysis by an expert. However, following further development, the method incorporates automated image analysis. Moreover, this application has an option for participants to edit any mistake related to segmentation, food labeling, and portion sizes .
Snap-n-Eat is a mobile food recognition system that uses machine learning algorithms to estimate nutrient intakes from photos taken by a participant . The algorithm behind this application is designed to distinguish foods in the photograph from its background without the user’s input, distinguish foods from each other using several features such as colour, texture, and size, and estimate portion sizes based on the number of pixels. The identification of foods is based on a training food dataset which is continuously expanding along with inputs of foods in new photos. The accuracy of the algorithm was found to be above 85% for the detection of 15 different foods.
This tool combines a series of 3 images of food consumed and also a short video . The combination of 3 images showed the highest accuracy of food classification and volume estimation only when one food is compared to several foods photographed at one time.
2.2 Dietary assessment with automated image-capture method
This method is a combination of automated image-capture with a web-based 24-hour dietary recall (e.g. Image-Diet Day  and SenseCam ). In this method a participant wears a lanyard around the neck. Connected to the mobile phone, every 10 or 20 seconds one image is captured, so it allows near-complete documentation of food and beverages consumed. Images passively taken by these methods help participants to better recall their food intake.
2.3 Smartphone applications for public health promotion
Apart from technologies developed solely for dietary assessment, there are also smartphone applications developed for public health promotion. Their main aim is to promote healthy dietary habits often in relation to weight management . While there is evidence of the effectiveness of such interventions in the short-term, studies are needed to also assess long-term effectiveness and sustainability .
Three applications to promote dietary behaviour change have been developed in Australia :
Innovative technologies for dietary assessment are used to overcome the problems of paper-based dietary assessment methods such as FFQ and 24-hour dietary recall. These limitations include:
There are studies showing that the acceptability of dietary assessment technologies can be varied across population groups with variable computer literacy, age, health, and sociodemographic status [49, 39]. Therefore, more and more studies are exploring the use of web- and mobile-based methods for collecting dietary intake data at the population level.
Population groups with variable cognitive skills and computer literacy
Young children, older adults and non-technology users are among those who may have less enthusiasm for engaging in the technology-assisted dietary assessment. A recent field study has suggested that these population subgroups may benefit from additional interview support [ 39]. However, the studies using a self-administered web-based 24-hour dietary assessment tool, Web-based Dietary Assessment Software for Children (WebDASC), in children (aged 8 to 11 years) have suggested acceptability and effectiveness [44, 35]. Using advanced features of mobile technology (e.g. receiving visual messages) may have impact on the response and accuracy among these population groups [16, 46, 49].
Dietary assessment among adolescents is also challenging because of irregular eating patterns and lack of personal incentives for recording dietary consumption. However this age group has reported to have higher acceptability for the technology-assisted methods compared to paper-based food records . Additional features can potentially be developed for this group, including tailored support of text or visual message with entertaining properties . In addition, an automated application for a mobile phone was studied to identify dietary consumption among adolescents (11-15-year olds). The method was assisted by a technology assisted dietary assessment tutorial video feature instructing how to capture images .
Populations with race/ethnic diversity
Dietary assessment in multi-ethnic population subgroups in a single study is challenging because of the complexity of their diets linked to diverse cultural background. The reason is that, for example, foods can be consumed in different ways such as sharing foods or eating with hands .
Dietary assessment studies targeting a population living in a low-income country remains limited due to the burden of high costs and complexity of data collection. A new project, The International Dietary Data Expansion (INDDEX), is aiming to address these issues by developing tools accounting for any culinary diversity.
If researchers intend to create their own software for the purpose of a study and not use a readily available one, there are several points to be addressed:
Although there has been a lot of progress in the area of technology-assisted dietary assessment, there are still some points to be considered. Measurement errors and bias are likely to be present because of self-reporting, possible instability of compliance, and possible limitation in computer literacy in certain population subgroups (e.g. elderly people). The methods require secure infrastructure for data transfer, other required systems, and related budget . In addition, smartphone applications for public health promotion have shown promising results from short-term studies, but long-term effects remain to be demonstrated .