Personalized and Culturally-Tailored Meal Planning Recommendation for Clinical Nutrition

DongHyeon Seo, Susan C. Kim, Keith Burghardt, Abigail L Horn

Keywords: Precision Nutrition, Expert-Guided Recommender Systems, Linear Programming, Semi-Supervised Learning

TL;DR: We introduce HumbleNutri, a meal prescription plan recommender system that generates personalized, culturally tailored meal plans for patients with specific dietary needs under the guidance of a Registered Dietitian Nutritionist (RDN).

Track: Findings

Abstract: Meal prescription interventions are essential for managing patients' dietary needs, yet existing approaches either require manual meal planning or rely on generic apps that lack cultural customization. We introduce HumbleNutri, a meal prescription plan recommender system that generates personalized, culturally tailored meal plans for patients with specific dietary needs under the guidance of a Registered Dietitian Nutritionist (RDN). HumbleNutri begins with a semi-supervised learning step to categorize recipes by meal type and cuisine, enabling culturally informed Medical Nutrition Therapy (MNT) recommendations. The system employs a modular framework that combines collaborative filtering-based recommenders with a bundle optimization model with constraints, suggesting meals that are aligned with patient preferences and MNT guidelines while ensuring that meal combinations satisfy patient-specific nutritional requirements based on their clinical profiles. Meals are organized into daily bundles (breakfast, lunch, dinner) and sequenced into weekly plans that support practical preparation and adherence to MNT targets. HumbleNutri translates clinical diet guidelines into culturally relevant meal plans, offering an equitable platform to deliver precision nutrition with an open-source toolkit and web application.

General Area: Applications and Practice

Specific Subject Areas: Deployment, Semi Supervised Learning, Representation Learning

Data And Code Availability: Yes

Ethics Board Approval: Yes

Entered Conflicts: I confirm the above

Anonymity: I confirm the above

Code URL: https://github.com/HumbleNutri/HumbleNutri

Submission Number: 201