Improving the accessibility and relevance of Medical Nutritional Therapy among people with Type 2 diabetes in the Latino community through a customizable, AI-powered application.

项目总结

PROJECT SUMMARY Medical Nutrition Therapy (MNT) is an effective treatment for managing Type 2 diabetes (T2D), but studies have shown that few T2D patients adhere to or access it. This low utilization rate can be attributed to the small number of registered dietitian nutritionists (RDNs) compared to the number of patients with T2D, limited reimbursement pathways for MNT, and resource constraints in the patient population. These issues are exacerbated in the Latino community due to limited access to care, economic disparities, and few resources tailored to culture identity. Currently, incorporating cultural relevance into an MNT plan must be done manually by an RDN. Existing digital applications focus on meal tracking or discrete biological factors (e.g. gut microbiome), supplementing healthy lifestyles rather than supporting MNT delivery and failing to address personal and cultural factors in meal plan creation and adherence. YumAI is a novel AI-based application that delivers and tracks adherence to MNT-compliant recipes that are customized to factors such as budget, family, location, season, food preferences, and preparation complexity. By automating meal plan adjustment, YumAI brings customized, high-quality MNT directly to patients through an easy-to-use digital application available on desktop computers and personal devices. In this Phase I SBIR project, we will further develop this AI-powered application, incorporating feedback from preliminary testing, ensuring compliance with ADA dietary guidelines and MNT stipulations, and optimizing the software specifically for individuals with T2D of Mexican decent. We will conduct two design thinking workshops to identify product features and discover factors in recipe customization and MNT compliance. Through an iterative design process, we will develop a proof of concept of YumAI and demonstrate algorithmic capabilities that enable basic recipe plan customization. Our two-step approach balances compliance with MNT standards and builds in relevant factors to customize recipe recommendations to user needs. Recognizing the complexity of plan customization and the number of possible dimensions of adherence, the purpose of the research is to demonstrate technical feasibility of a solution that can be enhanced in future development and to identify gaps in data collection. Successful completion of this project will result in a functional prototype of YumAI, the first digital health mechanism for increasing access to MNT for the Mexican population with T2D. This project will provide proof of concept and feasibility for future studies to scale YumAI for a broader population. It will prepare us for a Phase II project that explores in situ interaction with the product and efficacy of the product in solving issues of and patient adherence, cultural awareness, and RDN scalability.