AI nutrition recommendation using a deep generative model and ChatGPT

Area of Science:

• Nutrition Science
• Artificial Intelligence
• Computer Science

Background:

• AI in nutrition offers personalized recommendations but lacks expert guidelines, raising trust issues.
• Existing AI nutrition systems struggle with accuracy and trustworthiness due to a lack of established nutritional expert oversight.
• The integration of AI in personalized nutrition requires robust methods to ensure reliable dietary advice.

Purpose of the Study:

• To introduce a novel AI-based nutrition recommendation method.
• To enhance the accuracy and trustworthiness of AI-driven personalized dietary recommendations.
• To align AI-generated meal plans with established nutritional guidelines.

Main Methods:

• A deep generative network with sophisticated loss functions was employed to align with nutritional guidelines.
• A variational autoencoder modeled user anthropometrics and medical conditions in a latent space.
• An optimizer adjusted meal quantities based on user energy requirements, enhanced by ChatGPT for meal variety.

Main Results:

• The proposed method generated highly accurate, nutritious, and personalized weekly meal plans.
• Experiments on virtual and real user profiles demonstrated exceptional accuracy in meeting energy and nutritional requirements.
• The system showed increased meal variety, accuracy, and generalization capabilities.

Conclusions:

• The novel AI method effectively generates accurate and personalized weekly meal plans aligned with nutritional guidelines.
• The system demonstrates ease of integration into future personalized diet recommendation systems.
• This approach addresses the trustworthiness gap in AI-driven nutrition advice.