P642 - A CULTURALLY SPECIFIC PERSONALIZED MEAL PLANNER USING LARGE LANGUAGE MODELS FOR WEIGHT MANAGEMENT- A CASE STUDY ON A COHORT OF EGYPTIAN ADULTS

P642

A CULTURALLY SPECIFIC PERSONALIZED MEAL PLANNER USING LARGE LANGUAGE MODELS FOR WEIGHT MANAGEMENT- A CASE STUDY ON A COHORT OF EGYPTIAN ADULTS

M. Elattar¹, R. S. El Habibi², D. Z. Zaky³, M. E. Saber⁴, M. Abouelhoda^5,6, M. M. Hussien⁷, T. Mohannad^7,*

¹School of Information Technology and Computer Science, Nile University, Giza, ²Egyptian Aquatics Federation, ³Tropical Medicine, Ain Shams Univeristy, ⁴Research and Development Department, Intixel, Cairo, Egypt, ⁵King Faisal Specialist Hospital and Research Center, Riyhad, Saudi Arabia, ⁶Faculty of Engineering, Cairo University, Giza, Egypt, ⁷DigiTAAM FZCO, Dubai, United Arab Emirates

Rationale: Effective dietary planning plays a key role in weight management for obese individuals [1]. However, AI-based meal-planning tools often lack personalization, clinical grounding, and cultural sensitivity. This study introduces PMP-LLM, a modular, culturally aware system that leverages large language models and structured nutritional intelligence to generate individualized meal plans.

Methods: PMP-LLM is a dietary planning system built on detailed patient profiling, incorporating demographics, preferences, allergies, and cultural context. It combines a culturally aware food database, USDA validation, LLM-based meal generation, portion control, and safety checks. The study includes 378 healthy obese Egyptian adults (61% female, avg. age 34.2, BMI 33.8), with 22% reporting dietary restrictions and 13% allergies. Each received meal plans from PMP-LLM, a physician, and three commercial LLMs. Plans were evaluated for nutritional accuracy, USDA adherence, and cultural fit by two expert dietitians.

Results: Preliminary data from 378 participants show PMP-LLM outperformed commercial LLMs in macronutrient accuracy (92% vs. 79%) and USDA compliance (94% vs. 67%) while matching physician plans in most areas. Cultural compatibility scores were high for PMP-LLM (4.6), comparable to physician plans (4.7), and significantly higher than commercial LLMs (3.2).

Image:

Conclusion: PMP-LLM offers a novel AI-driven solution for culturally specific, personalized meal planning in weight management. By integrating safety, personalization, and cultural awareness, it delivers near-dietitian-level quality with enhanced scalability—making it a valuable tool for advancing individualized nutrition in diverse populations.

References: 1) Amiri, M.. Personalized Flexible Meal Planning for Individuals With Diet-Related Health Concerns: System Design and Feasibility Validation Study. JMIR Formative Research.

Disclosure of Interest: None declared