Evaluation of a ChatGPT-Based Personalized Prompt System for Diet and Exercise Planning in Adults with Overweight and Obesity: A Fuzzy Delphi Study

Artificial intelligence (AI) is gaining recognition for personalizing dietary and exercise plans to manage obesity. The effectiveness of AI-driven recommendations depends largely on the quality of input data and structured guidance. However, there is a gap in developing validated prompt-generation mechanisms specifically tailored for obesity-related lifestyle planning. This study aimed to assess and refine a personalized AI-driven framework (NExGEN-ChatGPT) for dietary and exercise prescriptions in obese adults, using the Fuzzy Delphi Method (FDM) to gather expert consensus. A panel of 21 experts from nutrition, medicine, psychology, fitness, and AI fields participated. Using structured questionnaires, the experts evaluated and refined six primary constructs, leading to the consensus validation of 111 specific criteria. The study identified key criteria for personalized, safe, and feasible AI-driven obesity management. In the “Goal Settings” construct, the ongoing assessment of physical activity and clearly defined target weight change goals achieved full consensus (100%) with fuzzy scores of 0.859 and 0.806, respectively. The “Demographics and Physical Characteristics” construct saw unanimous agreement (100%) on aligning recommendations with BMI categories (fuzzy score 0.681). In “Health and Medical History,” customizing plans for medical conditions also achieved 100% consensus (fuzzy score 0.683). For “Lifestyle and Behavior,” stress management strategies had 95.2% consensus and a fuzzy score of 0.784. In “Environmental Factors,” personalized shopping strategies reached a consensus of 90.5% (fuzzy score 0.768). Consequently, this validated framework provides a substantial foundation for subsequent real-world application and further research, thereby enhancing the effectiveness, scalability, and individualization of obesity interventions leveraging AI.