The Journal of nutrition · 2026
Comparison of ChatGPT and Dietitians in Formulating Diet Plans and Recommendations for Patients with Cardiometabolic Diseases.
СтатьяСравнительное исследованиеНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Katsigiannis George, Panoutsopoulos George, Perrea Anastasia, Detopoulou Paraskevi
Аннотация
BACKGROUND: Artificial intelligence is transforming personalized medicine, yet its efficacy constitutes a dynamic factor in the field of health and personalized medicine.
OBJECTIVES: This study aims to compare Chat Generative Pretrained Transformer's (ChatGPT) ability to construct dietetic plans and provide nutritional advice against professional dietitians.
METHODS: Three dietitians and ChatGPT-generated diet plans and gave recommendations for 1) obesity/dyslipidemia, 2) obesity/dyslipidemia/hypertension, and 3) obesity/dyslipidemia/type 2 diabetes, which were compared with each other and official recommendations. Prompts were developed via systematic iterative refinement. Macronutrient and micronutrient contents were analyzed using "Explore Food" software. ChatGPT's performance was also evaluated across sexes and 4 ethnic groups (Caucasian, Asian, African American, and Mexican). The nonparametric Mann-Whitney U test was used.
RESULTS: Dietitians recommended more carbohydrates, sugars, saturated fats, sodium, chloride, and iodine, whereas ChatGPT suggested higher polyunsaturated fats for obesity/dyslipidemia and obesity/dyslipidemia/hypertension. In addition, for obesity/dyslipidemia/hypertension, dietitians proposed more energy and fiber compared with ChatGPT. In the case of obesity/dyslipidemia/diabetes, dietitians proposed more sugar. Dietitians' plans were higher in vitamin C and other minerals in the case of obesity/dyslipidemia. Vitamin D was low in all plans. Accuracy relative to official guidelines was comparable: 50%-100% for dietitians compared with 55%-83% for ChatGPT across all conditions. No recommendation was made for adherence to the Mediterranean diet from ChatGPT, in contrast to dietitians. Overall, ChatGPT proposed higher energy plans for males than females (P < 0.05). However, ethnic subgroup analysis showed this difference was significant only for Caucasian and Mexican cases.
CONCLUSIONS: ChatGPT had a comparable ability to dietitians to design diet plans and provide nutritional counseling for people with obesity, hypertension, or type 2 diabetes. Its performance in specific ethnic groups may be limited. The value of human clinical judgment and interpersonal interaction in nutrition counseling is essential for patient engagement and optimal outcomes.
OBJECTIVES: This study aims to compare Chat Generative Pretrained Transformer's (ChatGPT) ability to construct dietetic plans and provide nutritional advice against professional dietitians.
METHODS: Three dietitians and ChatGPT-generated diet plans and gave recommendations for 1) obesity/dyslipidemia, 2) obesity/dyslipidemia/hypertension, and 3) obesity/dyslipidemia/type 2 diabetes, which were compared with each other and official recommendations. Prompts were developed via systematic iterative refinement. Macronutrient and micronutrient contents were analyzed using "Explore Food" software. ChatGPT's performance was also evaluated across sexes and 4 ethnic groups (Caucasian, Asian, African American, and Mexican). The nonparametric Mann-Whitney U test was used.
RESULTS: Dietitians recommended more carbohydrates, sugars, saturated fats, sodium, chloride, and iodine, whereas ChatGPT suggested higher polyunsaturated fats for obesity/dyslipidemia and obesity/dyslipidemia/hypertension. In addition, for obesity/dyslipidemia/hypertension, dietitians proposed more energy and fiber compared with ChatGPT. In the case of obesity/dyslipidemia/diabetes, dietitians proposed more sugar. Dietitians' plans were higher in vitamin C and other minerals in the case of obesity/dyslipidemia. Vitamin D was low in all plans. Accuracy relative to official guidelines was comparable: 50%-100% for dietitians compared with 55%-83% for ChatGPT across all conditions. No recommendation was made for adherence to the Mediterranean diet from ChatGPT, in contrast to dietitians. Overall, ChatGPT proposed higher energy plans for males than females (P < 0.05). However, ethnic subgroup analysis showed this difference was significant only for Caucasian and Mexican cases.
CONCLUSIONS: ChatGPT had a comparable ability to dietitians to design diet plans and provide nutritional counseling for people with obesity, hypertension, or type 2 diabetes. Its performance in specific ethnic groups may be limited. The value of human clinical judgment and interpersonal interaction in nutrition counseling is essential for patient engagement and optimal outcomes.