Digital dietetics and the era of artificial intelligence

We are in the midst of yet another technological revolution with the growing use of artificial intelligence and augmented reality. The challenge for dietitians and nutritionists is to discover how dietetics practice can effectively engage with these technologies and those from the preceding revolution. The beginning of this century saw the launch of social media websites sharing information, such as Myspace, soon followed by Facebook, Reddit, Twitter and Pinterest and now joined by Instagram and Snapchat applications (apps) for image and video sharing. In 2007, the iPhone was launched by Apple Inc and was transformational in how we communicate with each other, and after the advent of apps, has changed our professional, educational, financial and leisure behaviours. In this issue of Nutrition & Dietetics, there are articles from dietitians researching and employing communication and information technologies in their research and practice.1-5 The potential of new information and communication technologies in healthcare has been increasingly realised. Digital health modalities allow for real-time two-way communications or asynchronous one-way exchanges via electronic health (eHealth) using computers, or mobile health (mHealth) using smartphones.6 Wearable sensors and videoconferencing are increasingly used, sensors combine with apps to self-monitor blood glucose concentrations, blood pressure and heart rate to name but a few. Three papers in the current issue concern social media.1-3 Blogs are one of the simplest ways to engage in social media and are essentially a website where information is communicated, regularly updated and opinions can be shared rather than the more usual static informational website. Mete and colleagues have researched the profile of a typical Australian adult who reads healthy eating blogs.1 They investigated what readers of healthy blogs look for in a blog and why other individuals do not read them. Of the 151 healthy eating blog sites identified by the 238 participants, 27% of them were written by health professionals and proved most popular with participants. The authors suggest that blogs are still a useful medium for dietitians to communicate and encouragingly some participants reported they changed the way they eat on recommendations of these blogs. However, only half the survey sample read healthy eating blogs. The usefulness of social media in dietetics practice goes beyond the blog as illustrated by Squires and colleagues.2 Being aware that dietitians are using Instagram to post food and nutrition information, they created a checklist to assess the ethics and professionalism displayed in the Instagram posts of 50 dietitians and dietetics students. Of interest was that a substantial percentage of dietitians failed to indicate the source of their information, so it was unclear whether it was evidence-based. Additionally, any conflicts of interest were generally not considered by the dietitians, so few were posted. These are important considerations as stated in the Dietitians Australia guide published almost a decade ago.7 Social media can be a powerful means to promote healthy diets to Australians at no cost to the people reading it. The public engages with social media and Instagram is immensely popular. Research with young adults has demonstrated that 96% consider Instagram a suitable medium to disseminate food and nutrition information with videos preferred to text and images for engagement and motivation.8 Some dietitians may view social media as a means of ‘marketing’ to reach potential clients, but the ethics of the profession must be observed including declaration of conflicts of real and perceived interest. Any commercial interests such as industry-funded conferences or gifts or any sponsors or consultancies should be declared. Those dietitians who work in research and clinical trials know the value of using social media in recruitment and intervention. A group of paediatric obesity researchers led by Lang share their insights from using social media to recruit participants to dietary trials and create a useful checklist to follow.3 They talk about the ethics of social media in research recruitment and factors such as inclusion of vulnerable groups. Different forms of social media have also been used in research to deliver interventions. For example, Facebook has been used to deliver information, motivational messages, tips and recipes and allow interaction among intervention participants.9 There are many other ways that social media can be used in nutrition research, but interestingly few dietitian and nutritionists engage in exploring the big data available. Billions of people express their opinions on a plethora of nutrition issues. While the data can only be generalised to the population who make the posts, the data do offer insights into what people are eating and thinking about food. In a review of studies using social media to study dietary behaviours, 71 studies were identified from 2003 (inception of My Space) to 2021, with most studies published from 2016 onwards.10 It was found that only two of the studies included a researcher qualified in nutrition. Topics varied from analyses of a small number of posts to using methods to study millions of posts. Attitudes to certain foods such as organic foods and gluten-free foods were a feature. Both dietitians and the food industry may be able to gauge trends in acceptance of foods from such discussions. Studies also included assessment of food consumption patterns with the number of posts included ranging from the low hundreds to 50 million. Those studies with exceptionally large datasets used forms of artificial intelligence with automated computing programs and natural language processing software (a subset of artificial intelligence, machine learning and linguistics to interpret human communications). The smaller studies usually involve thematic analysis after posts are scraped from the web. So social media may provide a complementary dataset to traditional dietary surveys when trying to study dietary trends and attitudes. However, currently the ethical framework for dietitians and other health professionals for using the data is undefined and research should be submitted to an ethics committee. Users of social media may agree to the conditions of use without appreciating how their data may be used and that they can be sourced by third parties. Like social media usage, eHealth and mHealth are becoming part of everyday dietetics practice. The COVID-19 pandemic was a digital disruptor and moved many dietitians to deliver their services using digital technologies. While the telephone (smart or conventional) was an affordable easy option, videoconferencing for virtual consultations also gained in popularity. Increasingly in countries such as Australia, United Kingdom and the United States, virtual care is being funded and recommended by health service funders. The benefits for patients largely revolve around accessibility so that those in remote regions, with physical disability, and those who work in full-time employment receive convenient consultations with health professionals. The Dietitians Australia Position Paper on Telehealth published in 2020 reported that this medium was as effective as in-person medical nutrition therapy and that the emerging evidence base in eHealth and mHealth was positive.6 mHealth includes both text messaging interventions, apps and videoconferencing using apps. The Position Paper suggests that in-person, telephone and videoconference consultations may be enhanced by mHealth. One example is a randomised controlled trial using phone counselling and text messages among an overweight young adult population.11 Modest weight loss was achieved and maintained and dietary behaviours improved with the intervention group eating more fruit and vegetables, less take-away food and drinking less sugar-sweetened beverages. A systematic review and meta-analysis of the use of apps as an adjunct to usual care showed greater weight loss (Hedges g = −0.28; 95% CI −0.47 to −0.09) occurred when using apps.12 Further research into the use of mHealth as a stand-alone method rather than as part of a multicomponent intervention for the delivery of medical nutrition therapy is needed. Diverse groups of people may benefit from dietitians developing electronic and mHealth services. Wright and colleagues have undertaken exploratory research examining bariatric patients' barriers to accessing dietetics services and their preferences for digital delivery using mixed methods research.4 This study reports that text messages and emails were preferred modalities for information. In everyday medical nutrition therapy practice, apps and text messages may potentially save time and add economic benefit. An important consideration before recommending digital nutrition services is to determine the digital literacy of the individual as well as the resources available to them. Apps may be used across the nutrition care process such as for collecting information on patients' dietary intakes before the interview, goal setting and monitoring and supplying knowledge and skills information.13 Text message reminders are an effective means to improve attendance at healthcare appointments as confirmed in a systematic review.14 Electronic medical records enable prompt audits of patient records; machine learning can be used to program alerts to recognise patients at risk. Park and colleagues used electronic medical records for their study of nutrition in major trauma patients and outcomes.5 Before the advent of the electronic record, this type of study involved many hours going through paper records for relevant information. Although not explored in this issue, it is the burdensome task of dietary assessment where researchers have experienced gains this past decade. The relationship between diet, health and disease is dependent on valid assessment of food and nutrient intakes and tools for measuring food and beverage intakes are proliferating. We have automated 24-h dietary recalls that can be self-administered en masse for population surveys and a multitude of apps have been developed to replace the traditional food record. The advantage of these tools is that they simultaneously analyse the nutrient composition of the individual's dietary intake bypassing the process of dietitians entering all foods and amounts consumed into the analysis package. Nutrition researchers have designed and validated 24-h recalls and apps for prospective recording. There is still some burden on the individual recorder to ensure all meals, snacks and beverages are recorded, and caution should be exercised to ensure the participant can effectively use these apps. Some commercial apps have no dietitian and nutritionist input, and the source of the food composition data is questionable if crowd sourcing and commercial websites are utilised. However, if the source of nutrient data is confirmed, the portion size estimation is included, and the app has high usability scores, the tools are useful. Dietitians have worked on using smartphone digital images to record food intake for many years.15 While the burden is removed from the individual who needs to record their food and beverages, there are several barriers to achieving estimates of nutrient intake. Translating images into food and beverage portions is burdensome for the dietitian: portion sizes can be hard to estimate, some foods like mixed dishes are hard to discern, additions of salt, sugar and oils added to cooking are invisible and it is critical that the image is captured at the correct angle. Some of these barriers have been overcome with the use of machine learning systems (a subset form of artificial intelligence that uses algorithms trained on data) that are able to identify the foods and segment foods for portion size estimation. Wearables and sensors can be a useful addition to understand food consumption, revealing the when, where and how food is consumed as well as ‘what is consumed’ to researchers.16 Wearable cameras are not reliant on the participant taking a digital image.17 Cameras can take continuous images, so all consumption is captured and not reliant on the participant remembering to take an image at all eating and drinking occasions. Wearable cameras can be linked to an accelerometer or other systems that detect eating and turn the camera on and off that has advantages for analysis and participant privacy. The time and length of eating and drinking occasions can be monitored and the many environments where food is consumed, social settings of eating, and the concurrent use of screen technologies is captured. Lastly, we should further the role artificial intelligence may play in our profession. Artificial intelligence is differentiated from machine learning, and the subset of machine learning called deep learning, because it is meant to mimic human cognition, which includes reasoning and decision-making. Chat generative pre-trained transformer (ChatGPT) is a hot topic in education, research and business. It is trained on an enormous body of text data and uses deep learning to generate text resembling that written by a real person. The ethics of using copyrighted materials, such as books, and the use of this artificial intelligence in writing copy for business, school essays and publications, such as newspaper and research works, warrant urgent attention. Time will tell whether artificial intelligence will be used for clinical decision-making instead of dietitians.18 The best artificial intelligence chatbot may recognise human emotions but cannot develop its own emotions that are so important when we counsel patients. As with all technologies, dietitians will need to explore the benefits, beware the detriments, and develop ethics around responsible uses in their future practice. The author reports no conflict of interest for this editorial. She has received grant funding from NHMRC, ARC and NSW Health for unrelated work.