A Bilingual AI-Based Chatbot for Nutrition Education in a Food Is Medicine Intervention for High-Risk Pregnant Women: Design and Development Study

Top Things to Know

AI-based chatbots offer scalable, personalized support within FIM programs, but must be carefully designed to align with users’ needs, habits, and technological capabilities.

Factors like competing priorities, perceived relevance, and digital literacy strongly influence whether individuals engage with FIM supports.

Integrating human-centerd and cutrually tailored design into digital nutrition tools is critical to making FIM interventions effective for diverse, high-risk populations.

Summary of Conclusion/Findings

This study described the design, development, and iterative refinement of a bilingual AI-based nutrition chatbot (“Flora”) embedded within a Food is Medicine (FIM) intervention for high-risk pregnant women and evaluated its feasibility and early engagement patterns. Using a human-centered design approach and iterative plan-do-study-act cycles, the chatbot was tailored to deliver personalized, culturally relevant nutrition education via SMS in English and Spanish. Findings showed that integration of the chatbot into an FIM program was technically feasible, but user engagement was initially low, with 70% of participants not interacting and those who did engaging only briefly. Qualitative feedback revealed key barriers to use, including high cooking self-efficacy (reducing perceived need), low digital literacy, and competing life priorities that limited time and motivation to engage. Iterative refinements (i.e., reminders and visual guides) improved usability and cultural relevance but did not fully overcome engagement challenges, highlighting the need for greater personalization and integration into daily routines. Overall, the study demonstrates that while AI-driven nutrition education tools can be incorporated into FIM programs, sustained engagement depends on addressing behavioral, technological, and contextual barriers.