Enhancing Personalized Nutrition: Towards A Hybrid Intelligence Approach with LLM-Powered Meal Planning

Authors

Damette N., Tchappi I., Mualla Y., Markovich R., Najjar A., Abbas-Turki A.

Reference

Communications in Computer and Information Science, vol. 2471 CCIS, pp. 1-19, 2025

Description

This paper presents the LLM-powered Meal Planning Assistant, an innovative architecture designed to assist health professionals in creating personalized meal plans. This system leverages advanced Large Language Models (LLMs) like GPT-4o to interpret detailed user profiles and generate nutritionally balanced meal plans. The assistant demonstrates strong performance in accurately estimating caloric intake and reliably excluding restricted foods, underscoring its potential in personalized nutrition guidance. While the system performs effectively, there is room for improvement in the precision of macronutrient distribution, such as proteins and fats, which are critical for specific dietary requirements. Expert oversight is integral to the architecture, providing a crucial layer of verification to ensure the safety and accuracy of the recommendations. Additionally, the use of verified recipes datasets mitigates potential biases, ensuring reliable and safe dietary advice. This research highlights the potential of hybrid intelligence systems in advancing personalized nutrition, offering a scalable and adaptable solution for healthcare applications.

Link

doi:10.1007/978-3-031-89103-8_1