In a recent study published in the journal Dr nutrientsResearchers evaluated the potential of Chat Generative Pretrained Transformers (ChatGPT) to provide nutritional guidance.
Non-communicable diseases (NCDs) are the leading cause of death, accounting for 74% of deaths worldwide. The 2019 Global Burden of Disease study estimated that there were 43.8 million cases of type 2 diabetes (T2D), 1.2 billion cases of non-alcoholic fatty liver disease (NAFLD) and 18.5 million cases of hypertension. The prevalence of obesity nearly tripled between 1975 and 2016.
Various studies have consistently underscored the influence of lifestyle and dietary factors on the onset and progression of NCDs. Of late, internet searches for information on health related queries are on the rise. ChatGPT is a widely used chatbot that generates answers to text questions. It can understand context and provide coherent responses.
ChatGPT has emerged as an accessible and efficient resource for medical consultants. Chatbots can provide real-time, interactive, personalized patient education and support, helping to improve patient outcomes. However, data on the utility of ChatGPT to improve nutrition in NCD patients are limited.
Research and results
In the present study, researchers compared the nutritional advice given by ChatGPT with the recommendations of international guidelines in the context of NCDs. Analyzes were performed using the default ChatGPT model (version 3.5). This group included medical conditions requiring specific nutritional treatment, such as arterial hypertension, T2D, dyslipidemia, obesity, NAFLD, sarcopenia, and chronic kidney disease (CKD).
A set of prompts for these conditions, formulated by physicians and dieticians, were used to elicit dietary advice from the chatbot. Separate chat sessions were conducted for each prompt conversation. Responses to ChatGPT were compared with the recommendations of international clinical guidelines. Two dietitians independently assessed and graded responses to ChatGPT. Responses were considered “appropriate” if aligned with the guidelines and “inappropriate” if inconsistent
Additionally, responses were classified as “not supported” if they were not confirmed in the guidelines, “not fully matched” if they did not fully meet the guidelines, and “general advice” if they were non-specific and promoted a healthy diet overall. was done In addition, the team also evaluated whether ChatGPT could substitute consultation with a dietician in the management of complex cases and presented a scenario involving multiple coexisting conditions (CKD, obesity and T2D) in (one patient).
Overall, the advice given by ChatGPT was correct. Appropriateness rates ranged from 55.5% for sarcopenia and 73.3% for NAFLD. One response each for NAFLD and obesity contradicted the guidelines. Regarding obesity, the chatbot suggested regular meals and snacks to stabilize blood sugar levels, while the guidelines emphasized avoiding snacks between meals.
Regarding NAFLD, it has reported benefits from supplements such as omega-3 fatty acids, antioxidants and vitamin E under medical supervision. However, existing guidelines do not support antioxidants and omega-3 fatty acids for the treatment of NAFLD. A T2D-related recommendation was unsupported by the guidelines.
In particular, ChatGPT recommends dividing meals into smaller, balanced portions as an alternative to three large meals. Although this was not in direct conflict with the guidelines, existing T2D guidelines do not address it. Further, numerous responses were classified as not fully matching the guidelines.
For example, the ChatGPT emphasized the inclusion of foods with a low glycemic index and monitoring portion sizes for hypertriglyceridemia, while the guidelines recommend reducing excess body weight and carbohydrates. Also, ChatGPT offers generic health advice for several conditions. Examples include staying hydrated, avoiding processed foods, and including lean protein. Chatbots have repeatedly emphasized overall wellness and appetite management.
General advice includes foods that should be included in the diet, which guidelines often don’t report. Regarding obesity, CKD, and T2D patient situations, some recommendations from ChatGPT were inappropriate or conflicting. For example, it emphasizes prioritizing lean protein for muscle health and subsequently recommends limiting overall protein intake. Its responses were generally generic, repeatedly emphasizing consultation with a dietitian.
Findings highlight several points of agreement and variation among ChatGPT responses to dietary guidelines. The responses were clear and included practical examples of foods to include or exclude from the diet. Some of ChatGPT’s recommendations were partially complete. Chatbot fails to provide proper instructions in case of multiple co-existence conditions. Although ChatGPT was fairly accurate with regard to dietary advice for NCDs, limitations were evident in more complex situations. Thus, while ChatGPT may have potential utility, it cannot replace expert advice.