Smart Kiosk for Nutritional Management of People With Diabetes in Underserved Communities: Development and Technical Evaluation

Comprehensive Summary

This study, presented by Rivera-Garcia and colleagues, investigates the development and technical evaluation of an artificial intelligence–based interactive kiosk designed to provide personalized nutritional recommendations for individuals with diabetes in underserved communities. The researchers used a quantitative, experimental design to build and test a low-cost, offline-capable system that integrates a neural network trained on synthetic data derived from national clinical guidelines and local dietary patterns. The system was evaluated across three domains: accuracy of nutritional recommendations compared to expert-validated plans, system performance under varying user loads, and usability through heuristic evaluation and standardized scoring. The results show that the AI model achieved 87.3% overall accuracy, with high precision (90.5%) and sensitivity (92.1%) in generating appropriate dietary plans. The system demonstrated strong efficiency, with average response times of approximately 2.36 seconds, even under simulated concurrent use, and maintained stable computational performance. Usability testing yielded high satisfaction scores, with a System Usability Scale rating of approximately 89/100, indicating excellent usability.

Outcomes and Implications

Diabetes disproportionately affects underserved populations, where access to healthcare resources, nutritional education, and continuous monitoring is often limited. By providing a low-cost, offline, and culturally tailored solution, the study addresses critical barriers to equitable diabetes management and highlights the potential of AI-driven tools in community-based care. Clinically, the kiosk could support self-management and preventive care by offering personalized dietary guidance aligned with medical recommendations, potentially improving glycemic control and reducing complications. Its ability to function without internet access makes it especially relevant for rural or resource-limited settings, where traditional healthcare infrastructure is lacking. However, the study also underscores that the system is not a replacement for comprehensive care, as it currently focuses primarily on nutrition and does not integrate medication management or physical activity guidance.