This study is testing an artificial intelligence (AI) chatbot designed to help adults at risk for type 2 diabetes adopt healthier lifestyles. The chatbot provides personalized and culturally tailored guidance on nutrition, physical activity, resistance training, and goal setting. About 30 adults aged 18 to 55 years will use the chatbot for 6 weeks and provide feedback through surveys and interviews. Researchers will evaluate whether the chatbot is easy to use, helpful, and trustworthy, and whether data about changes in health behaviors can be feasibly collected, in order to inform larger future studies to improve diabetes prevention.
Type 2 diabetes continues to be a major public health concern, with growing evidence that individuals with normal body weight may also be at increased risk of developing prediabetes and type 2 diabetes. Existing diabetes prevention programs have primarily targeted individuals with overweight or obesity, creating a need for innovative prevention approaches tailored to younger, normal-weight, and culturally diverse populations. This study will evaluate a novel artificial intelligence (AI)-powered chatbot designed to deliver personalized, culturally tailored diabetes prevention support. The intervention combines evidence-based nutrition and physical activity guidance with conversational AI technology to provide accessible lifestyle coaching focused on diabetes risk reduction. The chatbot incorporates expert-validated educational content and behavioral support strategies intended to promote healthy habits and sustained engagement. The chatbot was developed using a curated knowledge base of culturally adapted diabetes prevention recommendations. Personalized guidance is generated using participant preferences, context, and validated educational resources to support lifestyle behaviors associated with improved metabolic health. The platform is intended to provide scalable, user-centered support while maintaining alignment with established diabetes prevention principles. This pilot study will assess the feasibility and acceptability of implementing an AI-enabled diabetes prevention intervention in a real-world setting. Researchers will evaluate user experiences with the chatbot, including engagement, satisfaction, trust, usability, and perceptions of cultural relevance. Findings will inform future development of digital health interventions and support the design of larger studies aimed at evaluating the effectiveness of AI-assisted approaches for diabetes prevention and health equity.
Study Type
INTERVENTIONAL
Allocation
NA
Purpose
PREVENTION
Masking
NONE
Enrollment
30
Participants will use a generative AI chatbot designed to provide personalized, evidence-based diabetes prevention support. The chatbot delivers culturally tailored nutrition and physical activity recommendations, with an emphasis on high-protein dietary strategies, resistance training, goal setting, motivational support, and healthy lifestyle behaviors. Participants will be encouraged to interact with the chatbot at least three times per week during the 6-week study period. Recommendations are generated using expert-validated diabetes prevention content and personalized according to participant characteristics and preferences.
Emory University
Atlanta, Georgia, United States
Feasibility and Acceptability: Task completion rate
The proportion of recommended chatbot activities, goals, or behavioral tasks completed by participants during the 6-week intervention period, as assessed through chatbot analytics. This measure will be used to evaluate participant engagement and adherence to chatbot-delivered recommendations.
Time frame: Baseline, 6 weeks
Feasibility and Acceptability: Conversation length
Average number of messages exchanged between participants and the AI chatbot during each interaction session over the study period. Conversation length will be obtained from chatbot analytics and used as an indicator of participant engagement with the intervention. Longer conversations may reflect greater interaction with chatbot-delivered nutrition, physical activity, and lifestyle coaching content
Time frame: Baseline, 6 weeks
Feasibility: Body Mass Index (BMI)
Proportion of participants for whom self-reported body mass index (BMI, kg/m²) is successfully collected at baseline and at the end of the 6-week intervention period. BMI will be calculated using participant-reported height and weight collected through study surveys.
Time frame: Baseline, 6 weeks
Feasibility: Physical Activity Behaviors
Proportion of participants for whom self-reported physical activity behaviors, including engagement in resistance training and overall activity levels, will be successfully collected through study questionnaires and chatbot interaction data.
Time frame: Baseline, 6 weeks
Feasibility: Dietary Behaviors
Proportion of participants for whom self-reported dietary intake and nutrition-related behaviors, including the adoption of higher-protein dietary practices and other healthy eating behaviors promoted by the chatbot intervention, are successfully collected. Data will be collected through participant surveys and chatbot interaction records.
Time frame: Baseline, 6 weeks
Feasibility: Behavioral Intention for Diabetes Prevention
Proportion of participants with successfully recorded self-reported intention and readiness to engage in healthy lifestyle behaviors for diabetes prevention, including dietary modifications and physical activity, as assessed through study questionnaires
Time frame: Baseline, 6 weeks
Perceived Trustworthiness of the Chatbot
Participant-reported trust in the accuracy, reliability, and credibility of information and recommendations provided by the AI chatbot, assessed through study questionnaires.
Time frame: Baseline, 6 weeks
Ease of Use: System Usability Scale (SUS)
The SUS is a 10-item questionnaire that assesses participants' perceived usability of the application. Each item is rated on a 5-point response scale. Responses are scored using the standard SUS scoring method to generate a total score ranging from 0 to 100, with higher scores indicating greater perceived usability.
Time frame: Baseline
NASA Task Load Index (NASA-TLX)
The NASA-TLX assesses participants' perceived cognitive workload while using the application across six domains: mental demand, physical demand, temporal demand, perceived performance, effort, and frustration. An overall workload score will be calculated from the domain ratings. Scores range from 0 to 100, with higher scores indicating greater perceived workload or task demand.
Time frame: Baseline
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.